I keep coming back to the Turtles because the system is almost embarrassingly simple, and yet it made a group of people with no trading background a lot of money. The rules fit on a napkin. Buy when price makes a new high over some lookback, sell when it makes a new low over a shorter one, size everything by volatility, and add to winners on the way up. That is it. No indicators stacked on indicators, no regime filter, no machine learning. So the interesting question is not whether it was clever. It clearly was for its time. The question is which pieces of it still hold up when you run them on markets that have gotten a lot more crowded and a lot faster.
What the original system actually said
The Turtle rules were built on Donchian channels, which are just the highest high and lowest low over a lookback window. A 20-day Donchian channel is the highest price of the last 20 days and the lowest price of the last 20 days. When price closes above the top of that channel, you are in a breakout. That is the whole entry signal. There were two systems running in parallel. System One used a 20-day breakout for entry and a 10-day breakout in the opposite direction for the exit. System Two used a 55-day breakout for entry and a 20-day breakout for the exit. Longer lookback, slower to trigger, harder to shake out.
The part people forget is the sizing, and the sizing is where most of the real intelligence lived. Position size was set in what they called units, and one unit was a fixed fraction of account equity divided by the market's volatility. Volatility was measured with ATR, the average true range over roughly 20 days, which is just the average size of a daily bar including gaps. A market that swings a lot gets a small position. A quiet market gets a bigger one. So a trade in a jumpy contract and a trade in a sleepy one both risk about the same dollar amount. That normalization is why the same rules could run across grains, currencies, metals, and bonds without blowing up on the wild ones.
Then there was the pyramid. As a winning trade moved in your favor by roughly half an ATR unit, you added another unit, up to about four units total. So a big trend did not just make you money on your starting position, it made you money on a position that grew as the move extended. The flip side is that a whipsaw could get you long the full stack right before it reversed, which is exactly as painful as it sounds.
Which parts decayed and which held up
Here is the honest read after watching versions of this run on modern markets. The short-term breakout is the piece that decayed the most. A 20-day channel break on a liquid crypto pair or a large-cap stock fires constantly, and a huge share of those breaks are noise that reverses within days. Everyone can see the same 20-day high. Market makers lean against it, momentum funds fade obvious levels, and you end up buying the exact tick where supply shows up. If you trade the raw 20-day system on liquid instruments today, you should expect a low win rate and a lot of small paper cuts. It is not that breakouts stopped mattering, it is that the obvious, fast ones got arbitraged into chop.
The long lookbacks held up much better. The 55-day breakout, and honestly windows even longer than that, still catch the kind of sustained regime change that trend following exists to capture. A market grinding to a genuine multi-month high is a different animal from one poking above last month's high. There is less to arbitrage because the position you have to hold to profit from it is uncomfortable and slow, and discomfort is not something the crowd rushes to compete away. Crypto in particular still throws off long, violent trends that a 55-day or 100-day breakout will sit inside for months.
The volatility sizing held up best of all, and if you take one thing from the Turtles this is it. Sizing by ATR is not a trend-following trick, it is just correct risk management. It keeps your dollar risk roughly constant as a market's volatility expands and contracts, which matters enormously in crypto where the same asset can triple its daily range in a week. Plenty of strategies that have nothing to do with breakouts get better simply by adopting ATR-based sizing.
An updated parameter set and what to expect
If I were handing someone a modernized version to paper trade before risking anything, it would look roughly like this. Treat these as starting points to test on your own data, not as gospel.
- Lengthen the entry. Use a 55-day breakout as the baseline and test out to 100 days. Drop the 20-day entry for liquid instruments, or demote it to a low-weight signal at most.
- Keep the exit tighter than the entry but not tiny. A 20-day trailing channel exit for a 55-day entry is a reasonable ratio. The exit is what protects the open profit, so err toward giving the trade room rather than snapping out on the first pullback.
- Keep ATR sizing exactly as designed. Risk a small, fixed fraction of equity per unit, one ATR-ish of stop distance per unit, and do not fudge it upward because a setup looks good.
- Cut the pyramid down or drop it. Adding up to four units is where the deepest drawdowns come from. One or two adds is plenty, and a lot of people are better off with a single entry and no pyramid at all.
- Add a long-term trend gate. Only take longs when price is above a slow moving average, something like a 100 or 200-day line, and only take shorts below it. This alone kills a lot of the worst countertrend whipsaws.
Now the part nobody likes. Trend following is a low win-rate way to make money, and this version is no exception. You should expect to be right well under half the time, with a string of small losses punctuated by a few large winners that carry the whole year. The drawdowns are deep and they last. A peak-to-trough decline of a third of your capital is not an anomaly for this style, it is a Tuesday, and the flat stretches between trends can run for many months while you keep paying small losses and feeling stupid. If that pattern would make you abandon the system at the worst possible moment, which is right before the trend that pays for everything, then the raw system is not the problem. The size you are running it at is.
Backtest it on your own instruments before you believe any of it, and pay attention to the shape of the equity curve, not just the final number. A system that ends up is easy to admire in a chart and very hard to actually hold through the ugly middle. The Turtles worked in part because the traders were handed the rules and told to follow them without arguing. Most of the edge left in this style is still exactly that, sizing it small enough and holding it long enough that you never have to argue with it.