Ick… disastrous finish. The last 100 hands were worse than the prior 150. I’m pretty confident this would be a winning strategy at this level, even as I played it, and later in this post (along with the prior post) I cover some things I think might help improve EV for the idea.
Sklansky Hammer results:
Tables played: 25/50 NL Holdem 9 max
Hands played: 1,002
Chips lost: -$2,129
BB/100 hands: -4.25
I was probably running better than 60 big blinds per 100 hands over the first 750 hands… so the final 250 hands constituted a pretty bad patch. I think I might have lost over a dozen shoves in a row at one point, getting chips in ahead on all but two of those. It’s not actually the case that I had even 50% equity on many of those, though, as quite a few were multi-way. On the two I was behind, my equity was quite poor, so maybe my average equity over that stretch was roughly 40% or even less (4 way, even with the best hand, 30% equity is pretty good). But even so, losing 12 in a row is not what I’d call running good. But so it goes.
Mechanics used:
For simplicity, base was just effective stack depth in big blinds. Since I was doing a min buy in of 50 big blinds and was usually the small stack, that meant for the first few hands around the table, my beginning key value was usually in the 40 to 50 range.
To modify that by the number of players I might face, I chose to generally weight all players that could still potentially call my all in bet equally, whether they limped or were left to act behind, and even often if they raised. Many players were limping with close to 100% of their hands, and so a limp may have made their hand weaker, just by reducing the likelihood of a premium holding. Even raises I often ignored, especially min raises from players that min raised pretty much every hand. If a given player did not limp often, or if there was a raise from a player that did not often raise, I weighted that player double. A 3! was usually counted as 4 (unless it was from a maniac that was always raising), and so 2 limpers, a raise, a 3!, and 1 player to act behind would be (2 + 2 + 4 + 1) for a total of 9. If I had 50 big blinds, that translated into a final key value of 450 (and so would have shoved with all of the hands through the 250+ row).
Hands shoved by key value:
4,000 +: AA
2,000 +: add KK
1,000 +: add QQ, AKs
500 +: add JJ, AK, AQs
250 +: add TT, AQ, AJs - ATs
100 +: add 99-88, SC → T9s, A5s - A3s
50 +: add 77, AJ, Axs, SC → 54s
25 +: 66-22, Ax, S1G
Again, I think the ranges in the last 3 are too wide (especially the last 2). I also speculated that the ranges might have been too tight on the upper most end, but I think that is a bit closer call. I got a stack of over 400 big blinds on a few occasions, and if I faced 6 or 7, even with no meaningful raises, that would push me over 2,000, where I would only shove with AA and KK, and I suspect QQ, JJ, AK and AQs are still profitable shoves as wide as people were calling.
I should also mention that I did flat a few times out of the small blind, and in those situations and as big blind, I allowed myself 3 options post flop: fold, call or shove.
Edit: I also think that people that try this but stick to the original Sklansky formula and shove ranges would see less variance than I did, on average. I wasn’t too worried about variance, but if you tried this with a bank roll of less than a million, you might consider that route.