I like powerful tools but black boxes make me a bit nervous, so in the hopes of understanding Commando's performance better I collected statistics on various entry signals and potential usage strategies. The results are very encouraging, with a few surprises.

Warnings and caveats

• For educational purposes, not financial advice.
• Past performance does not guarantee future performance.
• The data or code used to generate these stats may have errors. I did my best to sanity check results but guarantees are impossible. For what it's worth, I'm an ex-mathematician with training in data science. I'm happy to share the data and code used to generate these stats if there's interest.

Executive summary

• Commando scores as low as 1.5 are just as good or better than 2.0 and occur more often.
• Higher scores around 2.7-2.9 yield the best statistical edge. As long as Commando scores are increasing, the statistics are only getting more favorable. This ends once score reaches 3.0, but even then average outcomes are still better than when Commando crosses 2.0.
• High Commando scores yield especially good results for short term trades (8h to 24h), and work especially well with tighter stop losses.
• Trade management matters if one wants to get the most out of Commando. Using stop loss and take profit orders can improve outcomes, but not a huge amount over just hodling for a preset length of time.

The classic entry

Let's begin by looking at the classic Commando entry signal: a coin's score goes from below 2 to greater than or equal to 2. I looked at what happens over various time intervals after the entry signal. Below we have tables listing the averages, over all entry signals that fired for all coins since Commando's inception, of:

• max gain: the maximum percentage increase from the entry price over the time interval
• max drawdown: the maximum percentage decrease from the entry price over the time interval
• drawdown before peak: the maximum percentage decrease from the entry price before price reached whatever its maximum value was during the time interval
• hodl return: the percentage change in price from the entry signal to the end of the time interval

Note that this and all other data here exclude Commando signals for BTC because 1) most people use Commando to assess altcoins, and 2) the comparatively low volatility of bitcoin damps the gain/loss stats for altcoins that people are probably more interested in. I can share BTC specific stats if there's interest (they're just as good, but all the numbers are smaller because bitcoin price varies less than altcoins).

When interpreting these numbers, it's important to keep in mind what the average is taken over: for each entry signal we find the max gain, drawdown, etc over a certain length of time after that entry. That's one trial. What's reported is the average over all the trials. So, for example, the average maximum drawdown during 24h is -5.1%. This does not mean the maximum drawdown that ever happened was -5.1%, it means that after an entry signal fires one can on average expect a drawdown of -5.1% from the entry price sometime during the following 24h.

There is big variation here. Above are averages, but the data is widely spread around those averages. For example, the standard deviation for max gain after 24h is 6.2%, which implies that about 32% of the entry signals resulted in a maximum gain outside the range 0% - 12.3%. So there are big winners and also a few big losers. In 11.3% of cases price had increased 13% sometime within the 24 hours, but in 9.7% of cases price only moved downward in the 24h following the entry signal.