Emanuel Derman: Models. Theories. Reality.
Essential wisdom from the legendary quant and author of "My Life as a Quant" — understanding the limits of models for forex traders
Who is Emanuel Derman?
Emanuel Derman is a renowned quantitative analyst, physicist, and author, best known for his work on the Black-Scholes model at Goldman Sachs and his seminal book "My Life as a Quant: Reflections on Physics and Finance." He is a pioneer in the field of financial engineering and one of the most respected voices on the philosophy of financial modeling — a discipline he treats less as mathematics and more as a form of applied epistemology.
Derman began his career as a theoretical physicist at the University of Colorado and the University of Pennsylvania before transitioning to Wall Street in the 1980s, where he became a Managing Director at Goldman Sachs and head of the Quantitative Risk Management group. He is the co-developer (with Iraj Kani) of the Derman-Kani local volatility model — an extension of Black-Scholes that better captures how option implied volatility varies with strike and expiry. He has also written extensively on the limits of financial models, the dangers of mathematical overconfidence, and the fundamental differences between physics (where laws are immutable) and finance (where they aren't).
Born in 1945 in Cape Town, South Africa, Derman grew up in a Jewish family with deep intellectual roots. He earned his PhD in theoretical physics from Columbia University — a foundation that would later shape his distinctive voice in finance. At Goldman, he worked on the famous "Goldman Quant Bible" — an internal collection of model documentation that became legendary in the industry for its clarity about model assumptions. After leaving Goldman in 2002, Derman became a professor at Columbia University, where he directs the program in financial engineering. He writes a widely-read blog called Points & Figures, where he continues to publish essays on the philosophy of models, the limits of quantitative finance, and the role of narrative in markets.
What sets Derman apart from other quants is his willingness to publicly challenge the cult of the model. While most of his Wall Street peers built careers on selling the illusion of mathematical precision, Derman built his on explaining why that precision is mostly illusion. He coined (or popularized) the famous quote "All models are wrong, but some are useful" — a phrase originally from the statistician George Box, but which Derman turned into a foundational principle of the post-2008 quant culture.
For forex traders, Derman's insights are essential. He teaches that all models are wrong but some are useful — and knowing which are which is the true skill. His warnings about model risk, overfitting, and the illusion of precision are critical for anyone using quantitative tools in currency trading. A trader who takes Derman seriously will never confuse their backtest for the market — and that single insight can save a career.
From Physics to Finance: The Journey That Shaped a Philosophy
Derman's career began where finance ends: in the world of fundamental particles and immutable laws. That background — and the contrast with what he found on Wall Street — produced his distinctive voice on modeling, risk, and the philosophy of finance.
The Physicist's Mindset
Derman earned his PhD in theoretical physics at Columbia, where the goal of science is to discover universal laws that hold regardless of human opinion. He carried that epistemic standard into finance — and quickly discovered it didn't fit.
The Disorienting Shift
In physics, equations describe reality. In finance, they describe approximations of behavior shaped by human beliefs. The same math meant something fundamentally different — and Derman's writing is largely about helping people understand that gap.
The Writer's Voice
Derman's blog, books, and essays are the most accessible writing on quant finance ever produced. He has a rare gift for making epistemological ideas clear to traders, risk managers, and physicists alike.
Why This Matters for Forex Traders
Derman's physics-to-finance journey produced the most important mental model a quant trader can have: the map is not the territory. Your backtest is a story about the past, not a prediction about the future. Your indicator is a description of one slice of behavior, not a law of markets. Your risk model is a way to think about risk, not risk itself.
Forex traders who internalize Derman's philosophy build systems that survive regime changes. Those who don't — who treat their models as reality rather than maps — blow up. The choice is yours. The lesson is the same one Derman has been teaching for 30 years.
Career Timeline: From Theoretical Physics to Quant Philosophy
A condensed view of the milestones that shaped Emanuel Derman as a physicist, a quant, a quant leader, and one of the most important voices in financial philosophy.
1945 — Born in Cape Town, South Africa
Born in a Jewish family with deep intellectual roots. Grows up during apartheid era. Develops early fascination with how the universe works.
1970s — PhD in Theoretical Physics, Columbia
Earns his PhD and begins an academic career in particle physics at the University of Colorado and the University of Pennsylvania.
1985 — Joins Goldman Sachs
Transitions to Wall Street. Joins the quantitative strategies group. Quickly establishes himself as a leader in derivatives modeling.
1994 — Derman-Kani Local Volatility Model
Co-develops the Derman-Kani model with Iraj Kani, an influential extension of Black-Scholes. Becomes a standard tool in options market-making.
1990s–2002 — Head of Quant Risk at Goldman
Becomes head of the Quantitative Risk Management group. Authors the legendary "Goldman Quant Bible" — an internal collection documenting every model the firm used.
2004 — My Life as a Quant Published
Publishes his memoir, which becomes a classic of finance literature. The book opens with the famous line: "The models I worked with were mostly wrong."
2002–Today — Professor at Columbia
Joins Columbia's IEOR department. Directs the financial engineering program. Continues to write Points & Figures, his influential blog.
2011 — The Physics of Wall Street Published
Writes a second book on the history of quantitative finance. Continues to publish essays on epistemology, model risk, and the philosophy of markets.
Emanuel Derman's Philosophical Wisdom
Models, theories, reality — and the limits of all three in trading
"A model is a lie that helps you see the truth."
Models Simplify Reality
Derman's most famous insight: all models are simplifications. The art lies in knowing which simplifications are useful and which are dangerous. Every model is a map with deliberate omissions; the question is whether the omissions are aligned with the question you're trying to answer.
Forex Application:
- Use models (indicators, backtests) but understand their assumptions
- Never trust any model completely — always use multiple confirmations
- Remember: the map is not the territory; price action is the ultimate truth
- Document what your model ignores as carefully as what it includes
"In physics, you are trying to understand the world. In finance, you are trying to make money."
Finance Is Not Physics
Derman distinguishes between the immutable laws of physics and the human-driven, ever-changing rules of markets. Precision in finance is a dangerous illusion. The laws of finance are written by the same people they describe.
Forex Application:
- Don't treat backtest results as physical laws — markets change
- Be skeptical of strategies with "perfect" historical performance
- Accept uncertainty — probability, not certainty, is your tool
- Remember: 100% backtest accuracy is the surest sign of overfitting
"The purpose of a model is to illuminate, not to predict."
Illuminate, Don't Predict
Derman argues that models help you understand relationships and scenarios — not forecast the future. Prediction is hubris. The trader who treats their model as a flashlight, not an oracle, will use it more wisely.
Forex Application:
- Use indicators to understand market conditions, not to predict price
- Model different scenarios, then let price action confirm direction
- Avoid strategies that claim to predict exact price targets
- Reframe your model as a "thinking tool," not an "answer machine"
"The most important thing about a model is not what it includes, but what it leaves out."
Beware Omitted Variables
Every model simplifies by omission. The omitted factors are often where the biggest risks hide. The most famous crashes — 1987, 1998, 2008 — happened precisely in the dimensions the models didn't include.
Forex Application:
- Ask: what does your trading system ignore (sentiment, news, liquidity)?
- Build safeguards for black swan events that models can't capture
- Never rely on a single indicator — each has blind spots
- Build a "model risk dashboard" that tracks what could go wrong
"Having survived as a quant on Wall Street for 20 years, I learned that models are useful, but that financial markets are governed by the same forces as human nature — fear, greed, hope, and ignorance."
Markets Are Human
Derman's experience taught him that human emotions dominate mathematical elegance. Models fail when humans panic. The math doesn't capture what the market feels; the market doesn't behave the way the math says it should.
Forex Application:
- Blend quantitative analysis with behavioral sentiment indicators
- Expect model breakdowns during panic and euphoria
- Use wider stops and smaller size during extreme sentiment conditions
- Track COT data and retail sentiment as a "humanity overlay" on your model
"There is no such thing as a perfect model. There are only models that are useful for a particular purpose."
Fit-for-Purpose Models
A model that works for one market regime may fail in another. Derman emphasizes situational awareness over universal solutions. The right question isn't "is this model right?" but "is it right for what I'm trying to do right now?"
Forex Application:
- Use different indicators for trending vs. ranging markets
- Adjust your model parameters as market volatility changes
- Have a "regime filter" that tells you which model to use
- Track which model works in which regime — and switch accordingly
"The Greeks are not the reality. They are the partial derivatives of a model."
Greeks Are Model-Dependent
Derman warns that risk measures like delta, gamma, and vega are outputs of models, not objective realities. Over-reliance leads to false confidence. The Greek you see is the Greek your model wants you to see.
Forex Application:
- Don't over-optimize based on backtest statistics alone
- Understand that calculated "edge" is an estimate, not a guarantee
- Use model outputs as guides, not commands
- Always have a "manual override" for when the model is clearly wrong
"Quantitative finance is like trying to hit a target with a gun that moves unpredictably."
Markets Are Non-Stationary
Unlike physics experiments, markets change their behavior over time. Past relationships may not hold in the future. The target you aimed at yesterday may have moved by the time you pull the trigger today.
Forex Application:
- Regularly re-test your strategies on recent data only
- Be prepared to retire strategies that worked historically but no longer work
- Keep your models simple and robust to regime changes
- Add a "drift detector" that flags when live behavior diverges from backtest
"The only thing you can truly model is your own behavior."
Model Yourself First
Derman emphasizes that the most important model is the one that governs your own decisions and emotional responses to market events. You can't fix the market, but you can fix your reactions to it.
Forex Application:
- Create a personal trading rulebook that models your decision process
- Journal to understand your behavioral patterns under stress
- Build systems that protect you from your own emotional biases
- Treat your journal as a "behavioral backtest" — your most important one
"In physics, you can prove things. In finance, you can't. You can only persuade."
Finance Is Persuasion, Not Proof
Markets are driven by narratives and beliefs, not mathematical certainties. Accepting this uncertainty is liberating. The quant who demands proof will never find it; the trader who persuades others (and themselves) of a thesis can compound.
Forex Application:
- Treat your trading edge as a probabilistic advantage, not a guarantee
- Be open to changing your mind when new evidence appears
- Avoid absolute statements about market direction
- Build a "thesis in 3 sentences" before every trade
"The map is not the territory, and the model is not the market."
Map vs. Territory
Derman's core epistemological warning: never confuse your representation of reality with reality itself. The market always knows more than your model. The chart of an indicator is not the chart of price.
Forex Application:
- Always prioritize price action over model signals
- When model and market disagree, trust the market
- Use models as tools, not as masters of your trading decisions
- Build a "pre-mortem" check: "How could my model be lying to me here?"
"Model risk is the risk of being wrong in ways you haven't imagined."
Model Risk Is Unimaginable
The most dangerous risks are those your model didn't account for. Derman's warning echoes Taleb's black swan — but from a quant's perspective. Your model is most vulnerable in the scenarios it was never designed for.
Forex Application:
- Keep position sizes small enough to survive unknown risks
- Use stop-losses even when your model says "no risk"
- Stress-test your portfolio against historical black swans
- Build a "model risk" allocation — capital reserved for unknown unknowns
"The desire for certainty is the enemy of successful trading."
Embrace Uncertainty
Derman's physics background taught him to seek certainty, but finance taught him to accept its absence. Traders who crave certainty get crushed. The model that promises certainty is the most dangerous one.
Forex Application:
- Think in probabilities, not certainties
- Accept that being wrong 40-50% of the time is normal
- Focus on expectancy, not accuracy
- Refuse to use strategies that promise 90%+ win rates — they're lying
"The best you can do is to develop a few fundamental principles and then apply them with humility."
Principles Over Precision
After decades of modeling, Derman concludes that simple principles applied consistently outperform complex models executed inconsistently. The trader with 3 good rules beats the trader with 30 untested ones.
Forex Application:
- Master a few simple patterns rather than learning 100 indicators
- Apply your principles with discipline — that's your true edge
- Stay humble: the market can always teach you something new
- Cap your strategy at 3-5 rules; complexity past that adds risk
"Volatility is not the same as risk. Risk is the probability of permanent loss of capital."
Volatility ≠ Risk
Derman makes a crucial distinction: volatility is a statistical measure of price movement, but risk is the probability of permanent capital loss. Models that conflate the two can give a false sense of safety.
Forex Application:
- Distinguish between "feels risky" (high volatility) and "actually risky" (large loss potential)
- Size positions to the actual risk, not to the volatility number
- Low-volatility environments can be just as dangerous as high-vol ones
- Track permanent loss (ruin), not paper drawdown
"A good model should be like a good theory in physics: simple enough to be understood, general enough to be useful, and humble enough to be questioned."
The Three Tests of a Good Model
Derman distills the criteria for a useful model: simplicity (you can explain it), generality (it applies broadly), and humility (it admits its limits). Most quant models fail at least one of these tests.
Forex Application:
- Can you explain your strategy in 2 minutes? If not, simplify it
- Does your strategy work across multiple pairs and timeframes? If not, it's not general
- Build in a "kill switch" — a way to recognize when the model is broken
- Test your model against all three criteria quarterly
"The problem with quantitative models is not that they are wrong — it's that they can be wrong in a million different ways at once."
A Million Ways to Be Wrong
Derman observes that a complex model can fail in many subtle ways simultaneously. Simplicity isn't just elegance; it's risk reduction. The more parameters, the more places to break.
Forex Application:
- Prefer simple strategies to complex ones — fewer things to break
- Audit each parameter: does it add genuine value or just noise?
- Avoid strategies with more than 3-4 conditions; complexity kills
- Ask: "How many ways can this fail?" — keep that number low
"The plural of model is not necessarily wisdom."
More Models ≠ More Wisdom
Derman notes that running many models doesn't necessarily produce better decisions — it can produce paralysis, contradictions, and false confidence. Integration is harder than aggregation.
Forex Application:
- Run 2-3 models, not 10 — too many creates noise
- Build a clear hierarchy for when models conflict
- Don't add a new strategy unless it solves a real weakness
- Audit your "model zoo" quarterly — many will fail to earn their place
"A black swan doesn't care what your backtest said."
The Black Swan Problem
Derman agrees with Taleb: the most important market events are the ones your backtest never saw. The model that looks best in-sample is often the one most vulnerable to the unknown.
Forex Application:
- Reserve capital for unknown unknowns — keep a cash buffer
- Test your strategy against 2015 CHF, 2020 COVID, 2022 GBP crises
- Don't trust backtests that look perfect; they assume the future is the past
- Survive first, profit second — same logic at any scale
"Treat your backtest as a hypothesis, not a verdict."
Backtests Are Hypotheses
A backtest is a prediction, not a fact. The market will give the final verdict in real time. Derman reminds traders that the document most often confused with truth is the equity curve of an untested strategy.
Forex Application:
- Treat your backtest as a "probable future" not a "guaranteed future"
- Run paper trading for 1-3 months before going live
- Pre-write the conditions under which you'd abandon the strategy
- Build in expectation of decay — assume 20-30% performance hit in live
"A model that fits the past perfectly has explained everything and predicted nothing."
Perfect Fit, Zero Prediction
Derman warns that overfit models are precisely the ones that look most beautiful in backtest. They capture the noise of the past so thoroughly that they have nothing left to say about the future.
Forex Application:
- If your backtest looks too good, it probably is
- Pre-commit to: "Sharpe above 2.0 = suspicious"
- Test on out-of-sample data, never the in-sample fit
- Reward simplicity: if two strategies have similar backtests, take the simpler one
"The most important financial model is the one you don't see: your own behavior under uncertainty."
The Hidden Model
Every trader runs an implicit "model" of themselves: how they react to wins, losses, drawdowns, and uncertainty. Most traders never write this model down, which is why it surprises them in every crisis.
Forex Application:
- Write down your behavioral "model": your triggers, your reflexes, your patterns
- Stress-test yourself with paper trades during high-volatility events
- Identify your "drift moments" — when you start breaking rules
- Build rules that constrain your worst behavior, not just your best
"The most successful quants I know are the ones who can admit when their model is wrong — and do it quickly."
Speed of Admission
Derman argues that the speed with which you admit your model is wrong matters more than the model itself. The trader who kills a broken model in week 1 survives; the one who waits for month 3 doesn't.
Forex Application:
- Pre-write "kill conditions" — when to abandon a strategy
- When live results diverge from backtest for 30+ days, assume the model is broken
- Don't wait for the "definitive" signal — by then, you've lost 2x what you should have
- Reward yourself for killing strategies, not just for taking trades
"In markets, the question isn't whether you're right. It's how much you lose when you're wrong."
Loss When Wrong
Derman's late-career focus: the math of being wrong matters more than the frequency of being right. A system that's right 30% of the time and loses tiny on the other 70% beats a system that's right 70% of the time and loses big on the other 30%.
Forex Application:
- Optimize for "small loss on the wrong side" — that's the durable edge
- Cap each trade's risk at 1% of capital — no exceptions
- Track average loss vs. average win; the ratio matters more than win rate
- Use hard stops, not discretionary ones
"Once you've seen one financial crisis, you've seen one financial crisis. Each is unprecedented in its own way."
Crises Are Unique
Derman's late-career humility: even after 2008, the next crisis will surprise everyone. History rhymes but doesn't repeat. The most you can do is build a system that survives a wide range of surprises, not the specific one you imagine.
Forex Application:
- Build a portfolio that survives many possible crises, not the one you fear most
- Keep dry powder (20-30% cash) for unknown shocks
- Don't over-fit your risk model to a specific historical crisis
- Pre-define "any major crisis" rules: cut leverage, raise cash, reduce size
"A model is most useful when it tells you what you shouldn't do."
Models as Filters
Derman argues that the most valuable use of a model is elimination, not prediction. If your model says "don't trade this setup" — and you listen — you've gained as much as if it had told you to trade. The filter is the feature.
Forex Application:
- Use your model to eliminate bad trades, not just to find good ones
- Track the trades your model rejected — measure their hypothetical outcome
- Reward restraint: the trades you didn't take are often the most valuable
- Reframe "missed trade" as "saved from a bad trade"
The Derman Model Manifesto
Derman's foundational framework for thinking about financial models. Every quant trader who takes these five truths seriously will build more durable, more honest, and more useful systems. Master them — and your relationship to your own models will be forever changed.
Use models to understand relationships and scenarios, not to predict specific outcomes. They are flashlights, not oracles.
Accept that every simplification carries risk. Know what your model leaves out. Document omissions as carefully as inclusions.
Markets are driven by human behavior, not immutable laws. Precision is an illusion. Probability is the only honest tool.
The biggest risks are the ones your model doesn't capture. Prepare for the unknown with humility and capital reserves.
In finance, models can persuade but cannot prove. A model is a story about the world. Make yours a good one.
"After many years, I've come to believe that financial models are useful, not because they are true, but because they force you to be explicit about your assumptions."
The Derman-Kani Model: A Quant Landmark
The model Derman is most famous for, co-developed with Iraj Kani in 1994, is an extension of Black-Scholes that better fits observed option market behavior. While most retail forex traders won't use it directly, understanding its existence shows why Derman is considered one of the most thoughtful modelers in finance.
The Problem It Solved
Black-Scholes assumes volatility is constant — but traders know that volatility varies with strike and expiry (the "volatility smile"). Derman-Kani introduced a way to model this smile in a way that's consistent with the prices of all traded options. It was a major advance in derivatives modeling.
The Derman Philosophy in Action
The Derman-Kani model is a perfect example of Derman's philosophy: build a model that captures more of the real behavior, document its assumptions, and acknowledge its limits. It became standard in option market-making — not because it was "right," but because it was more useful than what came before.
Forex Analogy
Forex traders face a similar challenge: real volatility isn't constant. It varies with time of day, news events, and central bank cycles. A Derman-style forex model would explicitly account for these regime shifts — and document the limits of doing so. Most retail strategies don't.
Why It Matters
The Derman-Kani model illustrates the highest form of quantitative work: not "how do I optimize my backtest," but "how do I build a model that better matches the real world — and admit where it still falls short?" Most retail quant work skips step two entirely. Derman never did.
The Derman-Inspired Forex Playbook
Six concrete steps for applying Derman's philosophy of models to your own forex trading. None of them require a PhD — they require a willingness to be honest about what you don't know.
Document Your Model's Assumptions
Write down every assumption your strategy makes — and the assumptions you didn't realize you were making. If the assumption list has fewer than 10 items, you haven't thought hard enough.
Track What the Model Ignores
For every input you use, list the inputs you don't. Sentiment, news, liquidity, correlation — what would change if these moved? Build a "model blind spot" log and review it monthly.
Pre-Write the Kill Conditions
Derman is famous for asking "what would have to be true for this model to be wrong?" — and pre-committing to act on the answer. Your kill conditions should be specific, written, and bypass your ego.
Stress-Test Against Unknown Unknowns
Run your strategy against 2015 CHF, 2020 COVID, 2022 GBP — historical events that weren't in your backtest. If your model breaks, you have a model risk problem.
Reward Simplicity
If two strategies have similar backtests, prefer the simpler one. Simplicity isn't laziness — it's risk reduction. Complex models have more places to break.
Cultivate Epistemic Humility
The market is always more complex than your model of it. Read history, study the failures, and accept that you will be wrong often. The model is a flashlight, not an oracle.
The Derman Library: Three Books That Changed Quant Philosophy
Derman's book catalog is small but every title is foundational. Together they form the most accessible introduction to the philosophy of financial modeling ever written.
My Life as a Quant (2004)
Derman's memoir of his journey from theoretical physics in South Africa to the head of quantitative risk at Goldman Sachs. It is widely considered the best insider account of what it's actually like to be a quant on Wall Street. The opening line — "The models I worked with were mostly wrong" — set the tone for an entire post-2008 generation of quants.
The Physics of Wall Street (2011)
A short history of how physicists ended up dominating quantitative finance. From Bachelier to Black-Scholes to the 2008 crisis, Derman traces the ideas and people that brought physics tools to Wall Street. Required reading for anyone who wants to understand the cultural DNA of quant trading.
Points & Figures (Blog)
Derman's ongoing blog, hosted on Medium. He publishes essays on the philosophy of modeling, the limits of quantitative finance, the role of narrative in markets, and the meaning of life on Wall Street. Free to read. Among the most thoughtful voices in finance.
Build Models That Work — Without Being Fooled By Them
Learn how to develop robust quantitative systems, understand model limitations, and trade with probabilistic humility using Emanuel Derman's philosophical framework.
Deepen Your Quantitative Philosophy
My Life as a Quant
Emanuel Derman's classic memoir — the definitive account of life as a quantitative analyst on Wall Street and the philosophy of financial modeling.
Start learningUnderstanding Model Risk in Forex
Learn how to identify, quantify, and mitigate the risks that your trading models don't account for — Derman's framework applied to currency markets.
Explore toolsThe Philosophy of Systematic Trading
Combine quantitative rigor with epistemic humility — building robust systems that survive in non-stationary markets.
Get the frameworkFrequently Asked Questions
Common questions about Emanuel Derman, his books, and how to apply his ideas in forex.
Key Terms Glossary
Quick definitions for the quantitative finance, modeling, and physics terms used on this page.
Model
A mathematical or conceptual representation of a system. In finance, a model simplifies reality — it is not reality itself. Derman: "a model is a lie that helps you see the truth."
Model Risk
The risk of loss from using a model incorrectly. Includes flawed assumptions, misapplication, and failure to recognize the model's limits. The "risk of being wrong in ways you haven't imagined."
Black-Scholes Model
The foundational option pricing model from 1973, by Fischer Black, Myron Scholes, and Robert Merton. Assumes constant volatility — which is one of its main limits.
Local Volatility Model
An extension of Black-Scholes that lets volatility vary with strike and expiry. The Derman-Kani model is a famous example. Captures the volatility smile observed in real markets.
Volatility Smile
The empirical pattern that implied volatility varies with strike and expiry — out-of-the-money options trade at higher implied vol than at-the-money. Black-Scholes can't explain it; local vol models can.
The Greeks
Risk measures derived from option pricing models: delta, gamma, vega, theta, rho. Derman warns they are model outputs, not reality — relying on them is relying on the model.
Non-Stationarity
The property of financial markets that their statistical behavior changes over time. Past relationships (correlations, volatilities, regimes) don't persist. The biggest challenge to any model.
Overfitting
The statistical error of fitting a model so closely to historical data that it captures noise rather than signal. Produces beautiful backtests that fall apart in live trading.
"The world is not a model. The world is the world. A model is a story we tell ourselves about the world."