Table of Contents
Quantitative trader Michael Grech sits down for a candid interview about the real work behind going from manual chart-reading to fully and semi-automated systems. He explains how a decade of coding, testing, and iterating turned hard lessons—like early drawdowns, slippage shocks, and regime shifts—into a durable process. If you’re curious how a trader moves from picking value stocks after 2008 to building rules that run themselves, this conversation shows why Grech’s perspective matters for anyone who wants results without living at the screen.
In this piece, you’ll learn Michael Grech’s core playbook: keep strategies simple to avoid overfitting, respect costs (especially slippage) by trading less when needed, diversify across uncorrelated ideas and timeframes, and predefine a hard stop for strategy risk so you can survive the inevitable cold streaks. You’ll also see how he builds from observations to rules, backtests across long histories, and then lets automation enforce discipline—so emotions don’t hijack execution when markets change. By the end, you’ll have a clear, beginner-friendly template for turning a trading idea into a robust, rules-based strategy you can actually stick with.
Michael Grech Playbook & Strategy: How He Actually Trades
Core Philosophy: Simple Rules, Durable Edge
Here’s the big idea: keep the logic simple, let the math do the heavy lifting, and make sure the rules still work when the market gets weird. Michael Grech favors robust signals over clever ones, and he tests ideas across regimes so they don’t break the moment conditions change.
- Define every strategy in ≤5 logic blocks (signal, filter, entry, exit, risk).
- Prefer signals that work across assets/timeframes rather than niche optimizations.
- Require any edge to persist in out-of-sample periods and during at least one regime shift.
- If a rule needs heavy curve-fitting to look good, kill it.
- Cap model complexity: no more than 2 primary indicators and 1 confirmation filter per strategy.
Markets & Instruments: Liquidity First, Costs Matter
Before entries and exits, pick instruments that won’t tax your edge via slippage and fees. Michael Grech prioritizes deep, liquid products and trims size when spreads widen.
- Trade instruments with average daily dollar volume ≥ 20x your typical daily turnover.
- Hard rule: skip entries if current spread > 1.5× the 30-day median spread.
- Set per-instrument max slippage assumptions and bake them into backtests (use the larger of: modeled slippage or recent live slippage).
- Reduce target size by 50% during scheduled high-volatility events if the strategy isn’t specifically built for them.
Signal Construction: Objective, Repeatable Triggers
Signals should be easy to compute and hard to argue with. Grech leans on price-driven or simple statistical triggers that can be automated end-to-end.
- Use price/volatility inputs only (no discretionary pattern reads).
- Require signals to be computable once per bar (e.g., on close) to avoid lookahead.
- Minimum hold expectancy per strategy: positive after including fees and slippage at 2× your baseline assumption.
- Disallow overlapping, highly correlated signals in the same strategy; correlation > 0.7 triggers consolidation.
Entries: Let Price Invite You In
Entries are about alignment: only step in when your signal, the current regime, and liquidity are all on your side. That reduces churn and improves fill quality.
- Enter on bar close or at a single scheduled auction time; no intrabar chasing.
- Gate entries with a regime filter (e.g., trend or volatility regime) that must be true for ≥ 3 consecutive bars.
- If the signal fired in the last N bars (e.g., N=3) without profit ≥ 0.5×ATR, skip to avoid clustering.
- Use limit-at-close or next-bar limit with a tolerance band; market orders only during liquid hours.
Exits: Defined, Layered, and Unemotional
Exits are prewritten. Michael Grech stacks a profit-taking path with a hard stop and a time stop so no trade lingers and no outlier wrecks the month.
- Place an initial stop at k×ATR from entry (k typically 1.5–2.5 depending on asset).
- Stage profit targets: 0.75×ATR to de-risk (scale out 25–40%), 1.5×–2×ATR for core take-profit.
- Add a trailing stop that activates only after the first scale-out.
- Time stop: exit after M bars if neither stop nor target hit (M set per asset’s half-life of edge).
- All exits execute at bar close or auction times to reduce slippage surprises.
Risk & Position Sizing: Volatility-Based and Pre-Committed
Sizing is volatility-aware and capped at the portfolio level. The goal is to survive cold streaks without second-guessing the system.
- Size by risk per trade: risk_unit = (account_risk%) × (equity) / (stop_distance in $).
- Daily loss limit: stop trading new signals after portfolio drawdown = 1.25× your average down day.
- Max open risk across all positions ≤ 5–8% of equity (sum of per-trade stop risks).
- Reduce size by 50% after a 3-trade losing streak in the same strategy; restore after two winners or a flat month.
- No single instrument > 25% of total open risk.
Portfolio Construction: Spread the Edge, Not the Noise
Multiple uncorrelated strategies beat one “perfect” model. Grech blends trend-following, mean-reversion, and carry/term-structure style edges.
- Run at least two orthogonal strategies in each asset class (e.g., trend + mean-reversion).
- Cap strategy pairwise correlation at 0.5 over a rolling 6–12 month window; if exceeded, throttle the higher-volatility strategy.
- Target portfolio volatility (e.g., 10–12% annualized) and scale allocations monthly to that target.
- Add new strategies at a small weight (5–10%) and only scale after 6 months of live tracking.
Execution & Automation: Code Enforces Discipline
Automation removes hesitation and slippage born of indecision. Michael Grech codifies the full flow: signal → order → risk checks → broker.
- Pre-submit orders via bracket OCOs (entry + stop + target) where supported.
- Enforce pre-trade checks: spread, volume, earnings/events, and max risk limits.
- Log every decision point with timestamped inputs/outputs for post-trade forensics.
- If an order partially fills and the spread widens beyond the tolerance, cancel the remainder and re-queue the next bar.
Testing & Validation: Prove It Before You Trade It
A strategy isn’t real until it survives hostile testing. Grech stresses realistic fills, costs, and regime rotation.
- Split data into dev/in-sample, validation, and long out-of-sample; report each separately.
- Use conservative slippage (e.g., 1–2 ticks beyond historical average) and a fee buffer ≥ 20% above current.
- Stress test with randomized entry delays (1–3 bars) and widened spreads; the edge must remain > 0.
- Require performance across at least one crisis/regime break (e.g., crash, low-vol bubble) with no catastrophic drawdown.
- Kill strategies with profit concentrated in < 10% of trades or a single year.
Drawdown Control: Limits That Actually Trigger
Drawdowns are part of the game, but they’re bound by rules that activate automatically. That keeps the portfolio alive for the next opportunity set.
- Strategy kill-switch: disable and review if peak-to-trough exceeds backtest max DD by 1.2×.
- Portfolio circuit breaker: reduce all sizes by 50% after monthly DD > 6%, restore only after a +3% month.
- Lock last month’s profits by halving risk the following month if returns > 4% to avoid a give-back.
- Prohibit adding new strategies during a portfolio drawdown until a recovery threshold (e.g., +50% of DD) is reached.
Maintenance & Iteration: Slow, Documented Upgrades
Improvements are steady and measured. Michael Grech treats changes like code releases, minimizing unintended drift.
- Batch updates; never change more than one component (signal, filter, exit) at once.
- Shadow-trade revisions for 4–8 weeks before merging live.
- Re-opt small numeric parameters annually with tight bounds; never re-design the core logic to fit last year.
Data & Hygiene: Garbage In, Garbage Out
Clean data and consistent clocks prevent phantom edges. This is unglamorous work that pays for itself in fewer headaches.
- Normalize and checksum daily to catch missing bars, stale quotes, or split/roll errors.
- Pin timezone and bar close definitions; do not mix data vendors without reconciliation rules.
- Record the exact version of datasets and parameter files for each live trade so results are reproducible.
Psychology by Design: Pre-Commit So You Don’t Flinch
Discipline comes from systems that remove decisions in the heat of the moment. Grecch structures his day so that rules fire without debate.
- Fix review windows (e.g., 30 minutes after close) for decisions; no mid-day tinkering.
- Use automation to place orders and stops immediately after signals—no manual overrides except for tech failure or data error.
- Track a “rule adherence” score; if it drops below 95% for a week, reduce size by 25% until back above threshold.
Playbook for Different Regimes: One Engine, Two Gears
Markets swing between trending and mean-reverting. The portfolio shifts weights rather than rewriting the code.
- Maintain both a trend and a mean-reversion strategy per asset; tilt allocations monthly based on a simple regime metric (e.g., rolling autocorrelation or volatility ratio).
- During volatility spikes, widen stops by 25–40% and cut position sizes proportionally so risk per trade stays constant.
- Pause mean-reversion entries when intraday gap frequency exceeds a threshold (e.g., > 1.5× 6-month average).
Live Metrics That Matter: Fewer, Clearer KPIs
A small dashboard tells you if the engine is healthy. Michael Grech watches edge persistence, costs, and risk—daily.
- Track realized slippage vs. model; if live > model by 50% for two weeks, halve trade frequency.
- Monitor win rate × average payoff; deviations > 2 standard errors trigger a strategy review.
- Keep portfolio exposure heatmap by asset and strategy; rebalance if any cell > 30% of total risk.
Practical Daily Routine: What Happens Each Session
Consistency reduces errors. The daily checklist aligns the whole system with minimal attention.
- Pre-open: data integrity checks, event calendar scan, and spread/volume spot-checks.
- Post-close: compute signals, update orders, write audit log, and email a one-page summary to yourself.
- Weekly: reconcile broker fills vs. modeled fills and refresh the correlation matrix for strategy throttles.
Size Every Trade by Volatility, Not Gut Feel or Hope
Michael Grech keeps the size mechanical by tying every position to volatility. If a market is jumping, the position shrinks; if it’s calm, it grows within limits. That keeps dollar risk steady instead of letting emotions push bigger bets. It’s the difference between a process and a hunch.
Pick a fixed account risk per trade—say 0.5%—and measure volatility with ATR or standard deviation. Set your stop at a multiple of that volatility, then divide your dollar risk by the stop distance to get shares or contracts. Cap total portfolio risk and cut size after a losing streak, exactly how Michael Grech avoids compounding errors. By letting volatility, not confidence, dictate size, you stay in the game long enough for the edge to show.
Build Simple, Testable Rules; Let Automation Enforce Discipline Daily
Michael Grech keeps strategies brutally simple so they’re easy to test and hard to break. Each idea is reduced to a handful of if-then rules—signal, filter, entry, exit, risk—so performance doesn’t rely on clever curve-fitting. He stresses out-of-sample checks and regime testing, because a rule that only shines in one period is a liability, not an edge. If a component can’t be clearly stated or reproduced tomorrow, it doesn’t cut.
Then he lets automation police the plan so emotions never get a vote. Orders are staged with predefined stops and profit targets, and pre-trade checks block anything that violates spread, volume, or risk caps. Execution runs on a fixed schedule, logs every decision, and leaves no room for “just this once” overrides—exactly how Michael Grech maintains discipline through noisy days. The result is consistent behavior trade after trade, which is where small edges compound into real performance.
Diversify by Strategy, Underlying, and Holding Time to Smooth Drawdowns
Michael Grech spreads risk across uncorrelated engines so one rough patch doesn’t sink the month. That means pairing trend with mean reversion, carry with momentum, and letting each run on its own logic and cadence. He also diversifies by underlyings—equities, futures, FX, or ETFs—so the portfolio isn’t hostage to a single macro theme. Finally, he staggers holding times, mixing quick swings with multi-week holds to avoid synchronized whipsaws.
In practice, Michael Grech budgets risk, not capital, across these buckets and watches correlations like a hawk. If two strategies start moving together, he throttles the higher-volatility one until correlation cools. He caps any single instrument’s risk share and rotates weights monthly based on realized behavior, not opinions. The payoff is a smoother equity curve where losses are contained and wins compound without drama.
Respect Costs: Spread, Slippage, Fees—Trade Less When Liquidity Thins
Michael Grech treats trading costs like a line item that can erase edge if ignored. He checks the current spread against its recent median and simply skips signals when it’s too wide, because bad fills compound into ugly drawdowns. When conditions are borderline, he switches to limit-at-close or auction orders, letting the market come to him instead of paying up. If realized slippage runs hotter than modeled for a stretch, he automatically halves frequency until it cools.
He also sizes down when liquidity thins—smaller orders, slower pacing, and no chasing outside the most liquid hours. Michael Grech bakes fees and conservative slippage into his rule testing, so a strategy must win after all costs, not before. He tracks a live “cost drift” metric; if it creeps, he tightens the venue list or pauses the strategy. The rule is simple: protect the edge first, and only trade when the tape is cheap enough to make your math true.
Favor Mechanics Over Prediction; Define Risk, Stops, and Exit Timers
Michael Grech doesn’t try to guess the next headline; he enforces a repeatable script so every trade looks the same under pressure. That script starts with a fixed dollar risk per trade, a pre-calculated stop distance, and an order type chosen before the bell. If price hits the stop, he’s out without debate; if price reaches a partial target, he scales per plan and moves the stop only when rules say so. The goal is to convert market chaos into a checklist you can execute half-asleep.
He also time-boxes trades so dead money doesn’t linger—if nothing happens within a defined number of bars, the exit timer closes the position. Michael Grech keeps the language literal: trigger, confirm, enter, manage, exit—no “feel,” no “maybe,” no “wait and see.” When volatility spikes, he widens stops and cuts size proportionally so risk stays constant in dollars. By prioritizing mechanics over prediction, he turns small statistical edges into consistent results while protecting capital from impulse and regret.
Michael Grech’s core lesson is to turn trading into a rules-first machine and keep the logic simple enough that a computer can execute it without you. He moved from manual discretion to fully or semi-automated strategies, leaning on code to eliminate emotion and to backtest across long histories before committing real capital. In his words, computers only follow rules—so your edge must live in clear, repeatable instructions, not in feel.
He also hammers home that costs and regimes can make or break you. Early on, slippage and fast morning fills turned promising backtests into losses, which forced him to slow down, trade less frequently, and favor longer-term holds where costs don’t eat the edge. That shift pairs with an awareness of regime change: volatility, rates, and market structure can flip what works, so you need strategies that can sit out or adapt without constant tinkering.
Finally, Grech is strict about portfolio survival: predefine maximum pain, protect capital, and diversify by strategy and instrument so one engine can struggle while another carries the load. He advocates a hard, pre-decided cutoff for system risk—accept drawdowns as inevitable, but cap them and live to fight another day—and he broadened from stocks into ETFs and multiple strategy types to reduce running costs during tough periods. The mindset is pragmatic: when a strategy underperforms, fix the problem or pause it; when the market shifts, rely on simple, tested rules and risk limits to keep you in the game.
