Table of Contents
Kevin Davey sits down on the Desire To Trade podcast to unpack how a numbers-driven engineer became one of the most respected voices in algorithmic trading. In this interview, Kevin explains why he runs a large stable of strategies across futures markets, why simplicity beats cleverness, and how Occam’s Razor guides everything from entries to portfolio construction. He’s the author behind a widely read book on building algos and is candid about what actually holds up in live trading—versus what only looks pretty in a backtest.
In this piece, you’ll learn Kevin Davey’s end-to-end process: setting clear performance objectives before testing, keeping systems deliberately simple, validating ideas with walk-forward optimization, and letting new strategies “soak” on unseen live data before risking capital. We’ll also cover how he diversifies by market and timeframe, why he prefers straightforward dollar stops (and a smart hybrid with ATR caps), how he reduces stress by tracking equity weekly instead of daily, and why treating trading like a real business matters more than win rate. Expect practical rules you can copy, pitfalls to avoid, and a blueprint to scale from one idea to a portfolio of durable strategies.
Kevin Davey Playbook & Strategy: How He Actually Trades
How Kevin builds durable algos (the big picture)
Kevin Davey treats trading like an engineering problem: define the target, keep rules simple, validate hard, and only then deploy. The goal isn’t a perfect system—it’s a portfolio of good, robust systems that survive real markets.
- Set specific, measurable build goals before touching data (e.g., max drawdown, minimum trade count, realistic slippage/commissions included).
- Favor simple rule-sets over clever complexity; if two rules work similarly, pick the simpler one.
- Aim to create many “good enough” systems instead of searching for a single holy grail.
Strategy discovery and initial testing
This stage is about fast iteration without overfitting. Each candidate must prove basic viability across an adequate history and trades before deeper validation.
- Require at least several years of data and a meaningful number of trades per rule-set before you believe the results.
- Bake in slippage and commissions from day one, so “edge” isn’t just cheap fill assumptions.
- Kill ideas quickly if they fail base thresholds; move on to the next candidate.
Walk-forward and out-of-sample validation
Kevin’s core robustness check is walk-forward optimization (WFO) combined with out-of-sample testing to simulate real-time adaptation. Only systems that stay stable across multiple walks advance.
- Split data into rolling in-sample/out-of-sample windows; optimize only in-sample, then test on the unseen slice.
- Repeat across many walks; reject systems that only “work” in narrow windows.
- Keep parameter counts low; WFO magnifies the penalty for complexity.
Incubation: prove it live before risking capital
Even after passing WFO, new systems must “soak” live—tracked but not traded—for months. Incubation confirms behavior matches the backtest before real money.
- Paper-run each new system for 3–6 months and append results to the historical equity curve.
- Compare incubation equity to the WFO equity; if the shape diverges materially, shelve the system.
- Review incubation monthly; promote only those that “look” the same live as they did in testing.
Risk, stops, and risk-of-ruin control.l
The focus is on protecting the portfolio from tail events, not maximizing any single system. Kevin leans on simple stops and probability tools to size risk realistically.
- Use straightforward dollar or volatility-aware (e.g., ATR-based) exits to cap losses; avoid intricate, fragile exit logic.
- Run Monte Carlo on trade sequences to estimate drawdown distribution and risk of ruin before going live.
- Position size so the Monte Carlo “bad luck” path is survivable within your max drawdown tolerance.
Build a portfolio of many uncorrelated systems.
Kevin runs a large stable of strategies across markets and regimes, so losers are offset by winners. Diversification by market, timeframe, and logic is the edge multiplier.
- Target diversification across symbols (e.g., equities index, energies, metals, rates, FX) and across intraday/daily/weekly timeframes.
- Mix trend-following and mean-reversion ideas to balance regimes.
- Limit capital per system and per sector so no cluster can sink the ship.
Turn-on/turn-off discipline
Some traders toggle systems using equity-curve filters; Kevin’s lens is skeptical and evidence-driven. If you use filters, they must be pre-tested and rules-based—never discretionary.
- If employing equity-curve rules, define them in advance (e.g., MA cross of equity) and validate in WFO; no after-the-fact switches.
- Set clear “disable” triggers (e.g., breach of pre-defined performance band) and “re-enable” rules; log every change.
- Avoid micro-managing open trades; let systems execute by design.
Daily, weekly, and monthly operations
Execution is mostly automated; the real work is monitoring and periodic portfolio curation. Kevin advocates treating trading like a business with scheduled reviews.
- Automate signal generation and order placement where possible; keep a small manual subset only for platform-limited structures (e.g., some spreads).
- Dedicate a fixed review day each month to prune/replace weak systems and rebalance allocations.
- Keep a runbook: what to check pre-open, mid-day, pre-close; include latency, fills, rejected orders, and slippage vs. model.
Mindset that makes the process work
Kevin’s edge is as much temperament as technology: he ignores single trades, trusts the portfolio, and stays systematic. That mindset prevents the classic sabotage of turning systems on/off by feel.
- Judge systems by long-run equity behavior, not by today’s win or loss.
- Resist the urge to override signals; more systems make discretionary meddling exponentially harder and more harmful.
- Run trading like a business: documented process, measurable KPIs, and unemotional execution.
Proof that a process beats talent
Kevin’s track record was built on process: define, test, incubate, diversify, and execute. That repeatable structure—not a single magic indicator—delivers durability.
- Let results come from many small, independent edges rather than one oversized bet.
- Protect the downside first; upside takes care of itself when you survive the distribution’s worst paths.
- Keep building; the portfolio is a living product that improves one robust system at a time.
Size risk first: fixed-dollar and ATR caps that survive chaos
Kevin Davey starts every system by deciding the maximum pain he’s willing to take per trade, not by hunting for the perfect entry. He sets a fixed-dollar stop that reflects a small, repeatable slice of equity, then checks that the instrument’s normal volatility won’t blow through it in one routine wiggle. If the product is too jumpy, he scales the position down before anything else, because size—more than signal—decides whether you stay in the game. This mindset flips the usual script: protect the portfolio first, let the strategy prove itself second.
From there, he layers an ATR-based cap to keep rare volatility spikes from turning a normal loser into a catastrophe. If ATR expands, the allowed loss per contract stays constant, so the position size shrinks automatically and the stop distance widens only as much as the math permits. He also limits total exposure across correlated symbols so one market’s tantrum can’t hit five systems at once. By making fixed-dollar risk and ATR caps the default, Kevin forces every trade—no matter how tempting—to fit inside rules designed to survive chaos.
Build many simple systems; diversify markets, logic, and timeframes.
Kevin Davey builds a stable, not a superstar. Instead of chasing one perfect model, he stacks lots of simple, well-tested rules, each with a small but real edge. Fewer knobs means fewer ways to overfit, so he favors clean entries and straightforward exits that behave the same in live trading as they did on paper. The result is reliability through numbers: if one system drifts, the rest of the bench keeps the portfolio on track.
He diversifies in three dimensions at once—markets, logic, and timeframe. That means equity index, metals, rates, energies, and FX; trend-following, mean reversion, and breakouts; intraday, daily, and weekly rhythms. Kevin caps exposure by sector and avoids running five look-alikes that all die on the same day, watching rolling correlations so winners and losers don’t cluster. Allocation rotates toward uncorrelated performers, while laggards are pruned or replaced, keeping the ensemble fresh without betting the farm on any single idea.
Trust mechanics over prediction: rules, walk-forward tests, and incubation
Kevin Davey doesn’t guess where markets go next—he makes rules and tests whether they actually hold up. He uses walk-forward testing to optimize only on in-sample data, then immediately checks the rules on unseen periods to see if the edge persists. If performance collapses when the window shifts, the system is junk, no matter how pretty the backtest looks. He keeps parameters few and stable, so the same logic behaves consistently across multiple walk-forward runs.
Passing tests isn’t enough; Kevin incubates each system live for months before committing capital. Signals are tracked but not traded, confirming that fills, slippage, and equity shape match expectations in real time. Only systems whose live curve resembles their test curve graduate to real money; everything else gets fixed or shelved. That’s how mechanics—clear rules, staged validation, and patient incubation—beat prediction and keep Kevin in control.
Allocate by volatility and correlation, not gut feel or P&L
Kevin Davey sizes positions by how wild they are, not by how exciting the last trade felt. He targets a constant risk per trade and scales contracts so a high-volatility market gets less size than a quiet one. ATR or standard deviation drives the math, which means positions naturally shrink when volatility spikes and expand when conditions calm. He ignores recent P&L when allocating because winners can still be overcrowded and correlated, and yesterday’s hero can become tomorrow’s drag.
Kevin also watches correlati, so he doesn’t double-count risk across look-alike trades. He spreads capital across low-correlated markets and strategy types, and caps exposure per sector so one theme can’t dominate the book. Rolling correlation checks and a portfolio “heat” limit keep total risk inside a predefined envelope, regardless of how many signals fire. Rebalances are rules-based, nudging weight toward steadier engines while trimming clusters that start moving together.
Run trading like a business: review, prune, and redeploy capital.
Kevin Davey treats his trading like a real company with KPIs, audits, and scheduled maintenance. He reviews systems on a fixed cadence, not when emotions flare, comparing live results to their expected bands and checking slippage, latency, and error rates. If a strategy drifts from its performance envelope, it goes on a watchlist with clear, prewritten rules for fix, pause, or retire. This structure keeps him from reacting to single trades and lets the portfolio evolve deliberately.
He also runs capital like inventory: winners earn more shelf space, laggards get marked down, and dead stock is cleared. Kevin documents every change—when, why, and how much—so he can reverse course or learn from missteps. Cash freed from pruned systems is redeployed to uncorrelated performers or new, incubated ideas that just proved themselves. The net effect is compound discipline: steady reviews, clean books, and a portfolio that gets a little sharper each month.
Kevin Davey wins by engineering, not ego. He defines risk first, then fits every idea inside that box with simple, durable rules. Dollar-based stops are his default, often paired with an ATR-style cap so routine noise doesn’t knock him out, but panic spikes can’t wreck the account. He ignores the seduction of high win rates, judging systems by expectancy and risk-adjusted returns, not by how often they “feel” right. When conditions change, he re-optimizes on a schedule (every few months), treats that as maintenance—not magic—and keeps the logic lean so it continues to behave in live markets.
The real edge is scale and independence. Kevin runs a bench of straightforward strategies across futures markets, timeframes, and logics so that no single theme can sink him. He watches correlations, limits sector concentration, and redeploys capital into uncorrelated performers while retiring laggards. Operations are boring by design: roll contracts on time, check fills, track slippage, review systems on a fixed cadence, and document changes. It’s a business: clear KPIs, clean books, and zero room for discretion to hijack the plan.
If you’re copying one thing from Kevin Davey, copy the process. Start with the risk you can live with. Build simple rules that survive out-of-sample reality. Re-test and maintain on a timetable. Diversify by market, logic, and duration. And above all, measure success by portfolio health over months and years—not by today’s trade. That’s how you trade long enough for your edge to compound.
