Study: Risk Factors for Failure or Serious Trouble Are High Win % and Low Ratio of Avg Gain to Avg Loss

C2
1

The recent collapse of several of the top strategies at C2 allows us to try to figure out what are some of the risk factors for serious problems.

In this small study, I find that two risk factors for recent troubles are (1) a high win percentage and (2) a low ratio of average gains on winning trades v. average losses on losing trades.

I looked at the top 41 strategies by C2 score on the Grid on the afternoon of Feb. 5. While some strategies failed entirely this week (MCProTrader & U Got Mail), a larger number experienced (or continued) DDs of 20%-90%.

Fifteen of the 41 strategies—fully 36.6%–experienced major problems (greater than 20% DDs) during a week with very roughly a 10% closing price DD in stocks. Indeed, 4 of the top 5 strategies by C2 score had problems or failed.

image
image.png476Ă—559 17.6 KB

Note the mostly very high win rates and the comparatively low ratios of av. Gains to av. Losses.

  • Av Win %: 80.5%
  • Median Win %: 78.2%
  • Ratio Av Gain / Av Loss: 1.190
  • Median Ratio Av Gain / Av Loss: 1.442

Now compare the 26 strategies that did not experience major DDs, though some of them (including the top rated one, R Option) had more than a 10% DD.

[To see the entire Table, you will have to click on it.]

image
image.png577Ă—1097 74.7 KB

For the strategies with adequate performance, note the much lower win rates and much higher ratios of av. Gains to av. Losses.

  • Av Win %: 51.7%
  • Median Win %: 56.7%
  • Ratio Av Gain / Av Loss: 2.719
  • Median Ratio Av Gain / Av Loss: 3.150

Most of the troublesome strategies (60%) have win rates above 79%, while only 12% of the recently adequate strategies have win rates above that level.

Among the 41 top rated strategies on C2 on Feb. 5, those with a win rate above 79% had a 75% chance of failure or serious trouble in the last week, while those with a win rate of 79% or less had only a 21% chance of serious trouble.

Further, half of the troubled group had ratios of av. Gains to av. Losses of less than 1.2, and 36% (5 of 14) had ratios of less than .76. This compares with only 15% of the adequate group having ratios of less than 1.2 and only 1 of the 26 strategies in the adequate group having a Av Gain / Av Loss ratio below .76.

Putting this together for all 41 strategies, for the 7 strategies having both risk factors (win rate over 79% and ratio of less than 1.20), 5 of them (71%) either failed or had serious recent problems. For the 9 strategies with only one risk factor, 6 of them (67%) either failed or had serious recent problems. For the 25 top-ranked strategies that had neither risk factor, only 16% had serious problems in recent days.

RISK FACTORS: (1) High Win Rate & (2) Low Ratio of Av. Gain to Av. Loss

  • 2 Risk Factors: 71% Trouble Rate
  • 1 Risk Factor: 67% Trouble Rate
  • Neither Risk Factor: 16% Trouble Rate

CONCLUSIONS:

This small study tested two of the common hypotheses offered on this discussion list for system failure or serious trouble: high Win % and low Av. Gain v. Av. Loss. It examined the 41 strategies with the highest C2 scores on Feb. 5 and defined system trouble as ranging from a drawdown of at least 20% in recent days to complete failure (down over 100%).

Both hypotheses were supported by the data.

Nonetheless, this study looked at only one adverse event, and other times will be different. Further, neither of these two risk factors are guaranteed to generate problems for a developer. If a developer uses good money management, the risk of major problems could be very low, even lower than the strategies without these two risk factors.

Further, losing 20% in a week is something that probably most strategies at C2 have done. Yet if I had limited the analysis to comparing strategies losing more than 45% in recent days v. strategies doing better than a 5 or 10% recent loss, the relationships would have still been strong—and maybe even stronger.

Recommendations for Collective 2:

  1. Remove the % win rate from the Leaderboard, but retain it on the Grid. The default for that column on the Leaderboard should be a new statistic: Avg. Win / Avg. Loss .
  2. For each strategy page, remove the % Win Trades from the banner just below the strategy name. Again, replace it with Avg. Win / Avg. Loss .

Good luck trading.

Quantitative Models

12 Likes
2

Thank you for detailed analysis. Very much agree with it.

But, to new investors, or some that are very risk averse (oh the irony), high win rate gives them wrong sense of confidence. That is why high win rate systems still will be very popular, even after today.

3

The other two predictors for failure that are most commonly hypothesized here, and which I have not tested, are Martingale systems and Individual trade DD. -

4

Most martingale systems are covered by the high win % stat.

5

Most martingale systems are covered by the high win % stat.

Of course, you are right, almost all martingale systems are high win %, but not all high win rate systems are martingale…

6

Any high % win system has an issue even if its not martingale IMO. Either they are holding losers or adding to losers or otherwise not taking normal losses. Being able to take losses is critical. You either take down regular losses or big losses take you down.

3 Likes
7

For strategies above 90%, yes, but for strategies over 79%, usually, but not necessarily.

One strategy that has moved into the top C2 scorers is Payoff Matrix, with an 80% win rate. From what I’ve seen, it is probably safer than most of the other top strategies without the two risk factors. Of course, it’s a new strategy; if its model continues to work very well, but not quite as well, long term it might fall a bit below the relatively arbitrary 79% win rate that often spells trouble.

8

My personal cutoff for %win rates is 90%–that high or higher and you don’ t even get looked at. %Win 80%-89% and it’s a red flag, but I’ll at least take a look.

1 Like