Federer Won 80% His Matches — and Only 54% of His Points. That Gap Is the Lesson.
In his 2024 Dartmouth commencement address, Roger Federer told the graduating class he won almost 80 per cent of his 1,526 career singles matches. Then he asked how many of the individual points they thought he’d won.
Fifty-four per cent.
Toby raised that number himself when we spoke, and it’s the most clarifying thing in the interview. Short-term systematic trading can be the same kind of game. The best programmes don’t necessarily win most of their days by a wide margin — they tend to win only modestly more often than they lose, and they do it without emotional interference over an enormous sample. Elite performance in a competitive, efficient game rarely looks like clear dominance. It looks like a thin, relentless edge applied across a huge number of small, independent-looking bets.
Your equity curve behaves the same way. A hit rate tells you nothing about ordering, and ordering is what kills accounts. Losses arrive in convoys. Which is why Monte Carlo path simulation and loss-streak distribution matter more than the headline win rate, and why a backtest reported as a Sharpe figure with no path analysis isn’t telling you much.
Toby is candid that he doesn’t lose well and gets stubborn in his personal trading. His answer isn’t to become a better person. It’s to build the safeguards into the systems, so the threshold that says “this isn’t working” fires whether he likes it or not.
That’s what systematisation is for. Not precision. Protection from yourself.
But a thin edge only compounds while it’s still an edge. Which brings us to the uncomfortable part of the conversation.
Edges Die Slowly, and They Announce It First
“Everything is flux,” said Heraclitus, roughly 2,500 years before anyone thought to co-locate a server in Aurora, Illinois. Toby put it less poetically when I asked about the strategy that made his name: the damn thing’s had its worst years in the last three or four or five years.
Sit with that. The author of Day Trading with Short Term Price Patterns and Intraday Breakouts, the book that gave retail traders the ORB (Opening Range Break-Out), the NR4 (Narrow Range in 4) and the NR7, and which now sells for around $1,500 second hand, is the one telling you the trade has decayed. Meanwhile a generation is discovering the opening range on YouTube and treating it as a law of nature.
It isn’t. Maybe it never was.
Toby saw the first crack in 1986. The Board of Trade opened a three-hour evening session in the bond pit, a system called Project A. Traders suddenly had somewhere to unload risk overnight, the close-to-open pressure that powered the morning breakout had an escape valve, and follow-through dampened. Not to zero, and not overnight. It just got worse.
He could name the mechanism. That’s the part most traders miss. Toby didn’t watch an equity curve flatten and shrug, because he understood why the open carried information in the first place: with no overnight session, every participant on earth expressed their view in a single moment. The open was, in his phrase, a very precise area of energy.
Twenty-four-hour markets took that energy and smeared it across the clock.
So what’s left? A market taking out the previous day’s high used to keep going. Toby now puts that follow-through at 51 or 52 per cent, and suspects it’s lower, because institutional option flow means someone is standing in front of those moves all day long.
Read that against the number we opened with. A hair above half, applied relentlessly across thousands of trades, is a business. Fifty-two per cent — once you’ve paid the spread — is not. The entire distance between a thin edge that compounds and one that quietly bleeds can be a couple of percentage points, which should tell you how little room there is for a decaying edge.
The Principle Survives. The Parameter Doesn’t.
So has Crabel abandoned reference-point logic? Not remotely. He’s multiplied it.
In the early nineties he traded two markers: the opening range, and the previous day’s high or low. Today he counts fifteen or twenty in any given market, and rates the previous day’s close as more important than it has ever been. Time of day matters enormously, and differently in every market, because each has its own opening rituals, its own participants, its own cash market underneath. The Hang Seng’s fifteen-minute pre-open still produces something close to the clean momentum Toby remembers from the 1990s, which is why he rates the Asian session so highly right now.
The idea was never “buy eight ticks above the open.” It was “find where order flow concentrates, and trade the imbalance.” One is a parameter. The other is a research programme.
Traders who inherited the parameter are trading a corpse.
The Most Expensive Assumption in the Business
My favourite moment of the interview.
In the nineties, one of the largest short-term traders outside Monroe Trout ran three or four hundred million on a one-and-a-half to two-day momentum hold. Then he raised another three or four hundred. Toby asked how he planned to deploy it. The answer: extend the holding period to four or five days, so the extra size could be absorbed.
Toby told him the structure doesn’t work that way. Stretch from one day to five and the vibration in and out of the position grows far faster than the profit does. More volatility, more risk, and the slippage problem you were trying to solve reappears wearing a larger costume.
The firm was gone in three or four years.
There is no free capacity in a time frame extension. You pay for scale in execution engineering — Crabel has spent decades on tick-level execution research precisely to keep the cost of trading from swamping a short-term edge as size grows — or you pay for it in drawdown. Nobody skips the invoice.
What Toby Told Me to Tell You
I asked him directly what a solo systematic trader with a small account should do, given they can’t run three hundred markets and five hundred models. His answer was refreshingly unglamorous.
Start with one market. Take one basic idea, build multiple variations around it, and trade the ensemble rather than a single fragile configuration. Put strict rules around it. Decide in advance where the strategy isn’t working, and where the alpha isn’t there anymore. Then work outward: ten or twelve markets across a few regions gets you most of the diversification benefit if you’re automated.
And test through the shocks. Toby keeps intraday data back through 1987 for exactly this reason, and has pushed his ORB research to 1923, poor data quality and all. If your backtest has never seen October 1987 or March 2020, you don’t have a system. You have a fragile and pretty icon with a Sharpe ratio attached.
Key Takeaways
- Thin edges are normal, but losses cluster. The hit rate never tells you the ordering, so simulate the paths.
- Systematise your weaknesses. Toby knows he’s stubborn, so the exit threshold isn’t his to negotiate with.
- Decay is structural, not mystical. When an edge fades, name the market-structure change that caused it. If you can’t name the mechanism, you never understood the edge.
- Trade principles, not parameters. Reference points still work. The specific one Toby published in 1990 largely doesn’t.
- Time frame is not a capacity lever. Extending your hold to absorb size swaps a known cost for a much larger unknown one.
- Start with one market and strict rules. Then earn the right to add complexity.
The uncomfortable version of all this: the reason Toby is still here at seventy-one, still trading, still winning tennis tournaments in the over-65s, isn’t that he found the opening range breakout. It’s that he was willing to watch it die and keep researching anyway.