Hi-lo guide
Hades Hi-lo: cash out at step 2 or ride to step 5?
Do not read anything into "Hades"; it is a tag used to tell this page apart from its neighbours. The page compares two cash-out plans for the same hi-lo chain and shows what each one buys.
The two plans
Picture one fixed path of five calls. Plan A stops and collects after the second correct call. Plan B keeps going and collects only after the fifth. Both plans stake one unit at the start. In this model each call pays (1 - e) / p with a 1% edge, the next card is one of 13 equally likely ranks, and a tie loses.
| Step | Call | Step chance | Chain chance | Chain multiplier |
|---|---|---|---|---|
| 1 | 10 then lower | 69.23% | 69.23% | 1.43x |
| 2 | 5 then higher | 61.54% | 42.60% | 2.30x |
| 3 | Q then lower | 84.62% | 36.05% | 2.69x |
| 4 | 3 then higher | 76.92% | 27.73% | 3.46x |
| 5 | 8 then lower | 53.85% | 14.93% | 6.37x |
Side by side
| Measure | Plan A: 2 steps | Plan B: 5 steps |
|---|---|---|
| Chance to collect | 42.60% | 14.93% |
| Total return if collected | 2.30x | 6.37x |
| Expected return per unit | 0.9801 | 0.9510 |
| Expected loss per unit | 1.99% | 4.90% |
| Standard deviation of net result | 1.138 | 2.270 |
What the numbers say
Plan A collects more often and pays less. Plan B collects less often and pays more. That trade is the obvious part. The less obvious part is the expected loss row: Plan B costs about two and a half times as much as Plan A, because it pays the 1% edge five times instead of twice. Riding further does not only add risk; it also adds cost.
The spread row tells the rest. The standard deviation of Plan B is larger, so single results swing much further from the average. A player who prefers Plan B is buying a wider range of outcomes and paying a bit more for it. Neither plan has a positive expected value.
Why steps 3 and 4 matter so little to the payout
On this path, steps 3 and 4 are strong calls, lower from a Q and higher from a 3. They add only a factor of 1.506 to the multiplier, but they still pay the edge twice. Strong calls are the cheapest way to lose expected value without changing the outcome much.
Is there an ideal number of steps?
Not in terms of expected value. Each step lowers it by another factor of 0.99. Fewer steps always lose less on average.
What if I decide where to stop as I go?
The rule still holds. Whatever stopping rule you use, the expected return equals 0.99 to the power of the number of calls you actually make, averaged over the chains.