The idea before the name
Two plans both turn $10,000 into $13,000 over five years. The first climbs steadily, never falling more than a few hundred dollars below a previous high. The second gets there by doubling, halving, and clawing back, and at one point you were down $4,000 and seriously considering giving up.
They produced the same result. They are not the same plan, and anyone who has held the second one knows it. The question worth asking is: did this plan earn enough to justify how bumpy the ride was?
That question has a name, and the name is the Sharpe ratio. But the question is the useful part; the name is only shorthand so people can compare notes quickly.
How the number is built
Take what the plan earned over a year. Subtract what you could have earned doing nothing risky at all, such as holding cash in a savings account, because a plan that beats nothing is not paying you for the discomfort of holding it.
Then divide that by how much the plan bounced around, measured by how far its day to day results scattered from their own average. A plan that moves in small steady steps has a small number on the bottom; a plan that lurches has a large one.
The result is a ratio. Earning more moves it up; bouncing around more moves it down. That is the entire mechanic, and you never need to calculate it yourself.
A worked example with real numbers
Plan A earns 12% over a year. Cash would have earned 4%, so the part that was worth taking a risk for is 8%. Plan A's results scattered by about 10% over the year, so the ratio is 8 divided by 10, which is 0.8.
Plan B earns a more impressive 20%. Cash still earns 4%, so the risk-taking part is 16%. But Plan B lurched around by 32%, so the ratio is 16 divided by 32, which is 0.5.
Plan B made more and scores lower. That is not a flaw in the number, it is the whole point of it. Plan B paid you less per unit of discomfort, and if it had gone slightly differently the same lurching could have gone the other way.
What counts as a good number
As rough orientation: below 0.5 means you were not paid much for the discomfort you took on. Around 1.0 is respectable for an ordinary stock plan. Above 2.0 in a backtest should make you suspicious rather than pleased.
That suspicion is worth taking seriously. A very high number in a test usually means the plan was shaped to fit history unusually closely, or that the test period was unusually kind, or that something in the test was not honest about costs and fill prices.
Treat these bands as orientation and nothing more. A number of 0.9 versus 1.1 is not a meaningful difference and should never decide anything on its own.
Where the number misleads
It treats upward surprises and downward surprises identically. A plan that occasionally jumps up gets punished by this measure exactly as much as one that occasionally collapses, which does not match how anybody experiences their account.
It also flattens rare disasters. A plan that quietly earns a little every month for four years and then loses 40% in one week can show a comfortable number right up until the week that matters, because one catastrophe barely moves a four year average.
And it needs enough data to mean anything. Calculated over three months it is close to noise. This is why it belongs beside the worst drop and the trade count rather than standing alone.
How to use it in practice
Use it for comparing, not for judging. When you have two versions of your own plan tested on the same years, the one with the higher ratio gave you more return for the same discomfort, and that is genuinely useful information.
Never use it as a single verdict. In Stax it appears next to the worst drop in dollars, the number of trades, and the full trade list, because those together tell you something the ratio alone cannot: what happened, and whether you could have lived through it.