In 2008, the S&P 500 lost 38.5% of its value. Renaissance Technologies' Medallion Fund gained 98.2% that same year.
That's not a lucky year repeating itself. Between 1994 and 2014, Medallion averaged a 71.8% annual return before fees. Out of 144 months between January 1993 and April 2005, only 17 were losing months.
Almost none of the people running that fund would call themselves traders. Roughly half held PhDs in physics, math, or statistics, and Jim Simons was famous for saying Wall Street experience was frowned on internally.
Here's the part that gets skipped in every retelling of that story: Medallion doesn't have a strategy the way most people mean the word. It has a protocol.
A strategy is an idea. A protocol is a procedure.
A strategy is an idea about the market. Moving averages cross, so price will trend. Funding flips negative, so shorts are crowded. That's a hypothesis, and hypotheses are cheap — you can find a dozen of them in any trading forum by lunchtime.
A protocol is something else entirely. It's the documented, unambiguous procedure for testing that idea, deciding whether it clears the bar to risk real money, executing it, and shutting it off when it's wrong. Renaissance's models run largely on automation for exactly this reason: the firm removed the moment where a person could look at the output and decide they knew better.
Most retail traders have strategies. Almost none of them have protocols.
The edge people assume is not the real one
I think this is the real gap between institutional trading and everyone else, and it's not the one people assume. The common explanation is data, speed, or capital — institutions see the tape first, execute in microseconds, and have more leverage than retail ever will.
Those things matter, but they're not the core edge anymore. A retail trader on Bybit today has API access, decent historical data, and execution speed that would have counted as institutional-grade fifteen years ago. What that trader almost never has is the procedural layer sitting on top of the idea.
Watch what actually happens to a good strategy in the hands of a discretionary trader. The entry rule is sound. The backtest is real.
Then live trading starts, and every session becomes a negotiation: maybe skip this one, it doesn't feel right; maybe size up on this one, it looks obvious. Three months later the strategy hasn't changed at all, but the results are nothing like the backtest — because the thing that actually got traded was a strategy filtered through mood, not a protocol.
I've watched this happen with my own systems more than once. The strategy itself — a trend-structure entry on BTC — looks identical on paper whether that month goes well or badly.
What decides the outcome isn't the entry logic. It's whether I followed the protocol wrapped around it: the exact conditions that have to be true before an order goes out, the position size formula, the number of losses that ends my trading day.
The strategy is static. The protocol is the thing standing between me and overriding it in week three of a drawdown, which is exactly when every instinct says to change something.
Write it like a permission slip, not a description
Most people get this backwards because a protocol sounds like paperwork, and paperwork sounds boring next to a clever entry signal. So traders spend months refining the signal and never write down the procedure around it.
That's the wrong end to polish first. A mediocre strategy run through a strict protocol will usually beat a brilliant strategy run through none, because the brilliant one only shows up on the days the trader feels like showing up too.
If you want to test whether you actually have a protocol, try this: write your rules the way you'd write a permission slip, not a description. "I trade breakouts" is a strategy.
"I only enter if the 4-hour close is above the 50 EMA, price has retested it once, and my daily loss limit hasn't been hit — sized at 1% of equity, no exceptions" is a protocol. If any part of your rule still needs judgment to apply in the moment, it isn't a protocol yet. It's a strategy wearing a uniform.
This is really what automation buys you in systematic trading — not a smarter idea than the one you already have, but a protocol you can't argue yourself out of at 2 a.m. We publish our full rule sets and backtest data, including where they've failed, at v33systematic.com.