RA indexes outperformed by EV indexes

[hogwashap]

Using CVData, I can't find optimal indexes for Zen. Every time I generate indexes, I get slightly different numbers and I am unable to figure out what to do.

My RA indexes are also being outperformed in terms of SCORE by my EV indices somehow. For example, when I change my 10v10 EV index (+7) to the generated RA index (+11), I lose 0.17 (out of ~13.5 with my simplified sims for testing this). A little bit frustrated right now as I just don't know what to do or what indices to use.

Have been using CVData Beat-to-death sims with iRA multipass many times for maximum accuracy and can never outperform my EV indexes in SCORE. Both have the same indexes (sweet16 + fab4) but the SCORE change from switching to RA indexes is somehow negative.

Running 20 billion rounds each in CVCX to test results so fairly confident the indices are actually worse.

Any advice would be appreciated with how to proceed.

[hogwashap]

Update:
Changing doubling 10v10 from +7 to +11 is the only major loss somehow... I know it should be the biggest SCORE gain switching to RA. Everything else (9v7, 10vA is negligible and is probably within reasonable error (seems to be basically no change).

[DSchles]

You can't, of course, just change the index. You have to change the bet sizes, as well.

Don

[hogwashap]

You can't, of course, just change the index. You have to change the bet sizes, as well.

Don

Thanks for the insight Don. Would it not be the case that switching to RA indexes would increase SCORE and changing the bet sizing would allow us to keep RoR the same while increasing EV? Either way, I'm using optimal betting generated by CVCX so would that still be an issue?

[DSchles]

Thanks for the insight Don. Would it not be the case that switching to RA indexes would increase SCORE and changing the bet sizing would allow us to keep RoR the same while increasing EV? Either way, I'm using optimal betting generated by CVCX so would that still be an issue?

Reread BJA3, pp. 375-377. You can't increase SCORE by simply switching the indices. It's assumed that, when you do, the betting ramp will change accordingly. If, instead, you keep the same bets, then it's the risk that will be lowered.

Don

[Freightman]

Reread BJA3, pp. 375-377. You can't increase SCORE by simply switching the indices. It's assumed that, when you do, the betting ramp will change accordingly. If, instead, you keep the same bets, then it's the risk that will be lowered.

Don

To restate what Don has said except in a different way.
Theoretically, RA indices allow you to increase your max bet for the count in question. Practicality leaves you with the same max bet with that amount which theoretically would have been applied as an additional wager is now applied to reduce variance.

[Archvaldor]

Would it not be the case that switching to RA indexes would increase SCORE and changing the bet sizing would allow us to keep RoR the same while increasing EV?

Expected value goes down when you use risk-averse indices.

However, it is worth it because with risk-averse indices your variance is reduced.

In theory this should allow you to bet more aggressively for the same risk of ruin - meaning you make more money overall. If you don't or can't adjust your bet size, perhaps because it is impractical to do so in a real game, you will have lower EV and lower variance.

[Cacarulo]

Using CVData, I can't find optimal indexes for Zen. Every time I generate indexes, I get slightly different numbers and I am unable to figure out what to do.

My RA indexes are also being outperformed in terms of SCORE by my EV indices somehow. For example, when I change my 10v10 EV index (+7) to the generated RA index (+11), I lose 0.17 (out of ~13.5 with my simplified sims for testing this). A little bit frustrated right now as I just don't know what to do or what indices to use.

Have been using CVData Beat-to-death sims with iRA multipass many times for maximum accuracy and can never outperform my EV indexes in SCORE. Both have the same indexes (sweet16 + fab4) but the SCORE change from switching to RA indexes is somehow negative.

Running 20 billion rounds each in CVCX to test results so fairly confident the indices are actually worse.

Any advice would be appreciated with how to proceed.

The gain that can be obtained in Zen by using RA indices is not significant in shoe games. In 6D, S17, DOA, DAS, SPA1, SPL3, NS, 4.5/6 using EM indices [R22 + 10vT (+6)], spread 1:16,
the SCORE obtained is 27.35. If instead of EM indices, RA indices were used, the SCORE would be 27.59. As you can see, it's not very significant.
Now, if I only use the EM indices R22 but change the 10vT to +11, then the gain in SCORE would be 27.52.
If I directly remove 10vT (+6) from my repertoire, I would achieve a better SCORE (27.47) than by using it (27.35).

By the way, my EM index for 10vT is +6 instead of +7. The difference is due to the estimation of the remaining decks.
Hope this helps.

Sincerely,
Cac

[hogwashap]

Thanks Cac (and everyone else who replied)

I will trust your sims over mine that +11 is better than +6/7... Decreasing variance is much more important than increasing EV with my current bankroll. Appreciate the tips and advice everyone.

[Cacarulo]

Thanks Cac (and everyone else who replied)

I will trust your sims over mine that +11 is better than +6/7... Decreasing variance is much more important than increasing EV with my current bankroll. Appreciate the tips and advice everyone.

Regarding the simulations obtained with CVCX, I think you should increase the number of rounds to 100 billion. 20 billion for the minimum difference between EM and RA is too low.

Sincerely,
Cac