I'm trying to figure out what this HEM stat means and what I should learn from it.

Several searches in this and HEM forums have turned up very little.

Anyone know what this stat should look like e.g. 1.2, 2.6, 12?

If my RCE is 1.85 over 75k hands is that good/bad? should I be calling more or less?

Does it matter? If not, why not?

Any and all help is appreciated. Thanks in advance

Yours is fine.

This is a very, very hard stat to understand.

You should only look at this stat in extreme cases. Say you are 1 or 2.5 or so. then its probably a leak.

1.85 is fine.

I appreciate that. I want you to be right, of course, but would love to know why?

It's information that I have and would like to be able to use it/understand it.

Note: Based on the lack of information, I already assumed it is obscure and something deeper than 3bet% (which is pretty self-explanatory).

mine is 2.25, is this a leak ?

You can probably call in more marginal spots, perhaps.

Although you can't be sure of anything... since perhaps you just make a lot of money on your river calls(i mean perhaps you are just really good)?


I have tried to understand this stat but it is a bit hard. 2.25 you might want to look into your river calls or so, but like I said. You can't deduce anything from this.

It's too complex, like stats usually are. You have to break this down into several other stats to actually find a leak that you might have.

I would only look at this in extreme cases.

I think the definition of #2 is wrong

The definition of RCE is suppose to be such that RCE of 1.0 means your river calls are breaking even, a value above 1.0 is making money and a value below 1.0 means your river calls are losing money.

For this definition, #2 should be

2) If you lose then it 0 and if you win then it it the total amount won.

With this definition, RCE can range from 0.0 (every time you call the river you lose) to <really big> (villain is always min betting the river, you call and always win).

If you want 0.0 to be the break even point for RCE then a call and loss is still 0 but a call and win ==> pot - call size. OTOH If you make 0.0 the break even then the stat should be called river call return on investment, not river call efficiency.

In the world of physics, energy conversions always result in losses but a perfect lossless conversion would be called 100% efficient. I would assume the creator of RCE is familiar with science nomenclature and therefore wants RCE of 1.0 (100%) to represent break even.

Heh, I am still not understanding the posts in this thread written as rules and conditions and such like.

Here is what I do know: The river call efficiency stat answers the question: For every dollar I call or check/call on the river, how much money do I win, on average?

So a RCE of 1.85 is telling you win $1.85 for every dollar you call off on the river.

(This is a solid RCE, btw, and it means you are probably winning about 41 or 42% of the time you call the river).

River call efficiency goes down to 0. I had a filter on the other day that filtered for pots lost and I noticed that the RCE displayed for that subset of hands as 0.

After thinking about this more and given that the first stat is named River Call Net$ , HEM is probably using the "break even => RCE 0.0" formula. Physics nomenclature not withstanding. I.e. I think it would be more correct name would be River Call ROI since ROI = 0 traditionally means break even.

Assuming we know which RCE value means break even (0.0 or 1.0 depending on your definition. 0.0 in HEM's case) the meaning of the stat is pretty clear IMO. I would expect varying river call frequency (assume all other factors are held constant) to result in a laffer curve for River Call Net$ over a fixed sample size.

Presumably optimal RCE depends on playing style. I would expect an optimal RCE for a LAG to be higher than for a TAG but that is just a guess.

Mpethy, any comments on which RCE values (HEM's definition) you think are optimal and whether the optimal value varies according to playing style?

Depending on the sample size I look over, I have an RCE of +0.7 (large sample including tilting sessions with showdown monkey behavior) to +1.2 (small sample with fairly nittish play).

Oh, in the original HEM thread IFORGET is confused by the fact that you can be losing money overall but have a positive RCE.

This is actually easy to explain: you play horribly PF, on the flop and turn but make great river decisions e.g. weak-loose who insists on seeing the river but makes good decisions on that final street.

In RCE, the stat only counts the money you lose on the river call against you. All the money you dumped into the pot on earlier streets is "profit" when you win it on the river and counts as 0 loss when you fold the river.

TANGENT: the great thing about analyzing river play (river call in particular) is that it is the mathematically simplest as a river call (in an HU pot) is the last action before showdown so analysis of our action does need to be controlled for future actions.

I'm not sure that optimal river call efficiency relates to playing style. presumably we are all making the same calculation; if it is profitable to call, we call. RCE will be affected by playing style to a certain extent, of course. But I have not really noticed a strong relationship between style and RCE.

I don't have an opinion what RCE is optimal. The vast majority of winning players I see are between 1.7 and 1.95 The best player I have ever dne a db analysis for is also the person who had the highest RCE I have ever seen--well over 2.0. That was one of those things that makes you go, "hmmm." So I have started to look at RCEs below 1.7 as a potential leak. Normally if I see it down there, I look at some other river stats and will look at some hands to see if the person is calling down unreasonably light.

Your RCE makes me think, "holy ****, he is callin a lot of rivers." If you also have a high WTSD (>26%) and a low W$SD (<52%) and a River call win% around 35%, my strong suspicion would be that we are looking at a leak. If any of those 4 stats are different for you than what I have listed, I'm not sure what I would think right now. I'd want to dig more, but I'm not sure where--maybe in your river bet stats.

I don't use HEM, but the way I understand this statistic it tells us how much we get back for any dollar we put into the pot calling or check-calling the river. A priori I can't see an ideal value. The only thing that's seems clear is that by folding every hand we break even, which is equivalent to RCE=1. So we need to do better than that.

Suppose we face a PSB. If we have the nuts we always win the pot which contains three times our bet, so we achieve RCE=3. If we call with the anti-nuts of course we get RCE=0. However, we should call with each hand that has a probability of winning of at least 1/3. Thus the profitable calls range between RCE=1 and RCE=3. The average RCE of all profitable calls will lie somewhere in the middle, probably in the vicinity of 2 (or a bit lower, assuming that nut hands are rarer than good enough hands).

The range of RCE for profitable calls changes with the size of the bet. For a half-pot bet a call with the nuts has RCE=4; a third-pot bet gives RCE=5, etc.

Ideally we make all the profitable calls, and none of the unprofitable ones (which would contribute values less than 1.0 to the overall RCE). Since the average bet size on the river is probably somewhere between half pot and full pot I'd guess a good value would be somewhere around 2.25 (halfway between 1 and (halfway between 3 and 4)) or a bit lower. A value that is significantly higher would indicate that the player is too nitty with his river calls. A lower value means that some calls are unprofitable.

That's just the musings of a hobby theoretician. I hope I have expressed my thoughts relatively clearly. I personally am not yet sure as to the usefulness of this statistic.

Have you considered that if we make a call on the river getting 2 to 1 against a range where we are beating 50%, this is a correct call even if we lose? In other words, whether we win or lose the pot has no bearing on whether it was a profitable call, but does have a bearing on our RCE.

If you follow the logic of river calling and RCE, we don't want a number that is too high. We want to be making marginal river calls that will result in a somewhat low RCE.

Just rambling, no real point.

Yes, I considered that. What I consider a profitable call may lose money in the actual hand.

Edit: Yes, there will be marginal calls. There will also be calls with the nuts. I tried to give a handwaving proof that the average RCE for profitable calls will lie above 2.0.

Clarification:

Yes, but over the long run profit and theoretical profitability will converge.

This is basically how I use it, but I think 2.25 is too high for "good." The highest RCE I have ever seen above the low micros was 2.12, and, as I said, most good players are between 1.75 and 1.95.

So if I see a value outside this range, I basically investigate to see if there is a problem.

So I'm on the high side: 1,95 over the last 100k at 50NL and 2,2 at 25NL over the last 120k (to be exact: 2,03 over 80k at the $ tables and 2,41 over 40k hands at the eurotables).

What kind of leaks are correlating with a high RCE like mine?

I'm actually pretty surprised, cause I feel like a callingstation and often making light calls on the river. However: I tent to bet for value on the river way wider than most people and rarely choose to c/c when OOP. Often I either bet for value or c/f unless villan keeps it small (as I always feel that villans calling range on the river is way wider than his betting range). So thinking about it, maybe it's a leak that I'm missing good c/c spots where I can induce bluffs, instead of having villan fold for my 3th valuebet?

The value I came up with is too high because a) mediocre hands are more common than nut hands and b) we rarely c/c the nuts.

I'm cooking up the stat in PT3 myself (with help from PT3 support). I think I've successfully defined the "1.0 is break even" version of the stat. I just tried to make the "0.0 is break even" that HEM apparently uses and it is going haywire.

I think I should simply be able to subtract 1.0 from my "1.0 is break even" version of the stat to get HEM's version.

In that case, my HEM RCE is the 0.7 to 1.2 quoted earlier.

Strangely enough the stats (PT3) you ask about look OK:

• WTSD: 25.5%
• W$SD: 54.4%
• W$SD after R call: 43.4%

(not great but OK). But yeah, I've always thought (intuitively) that being a showdown monkey is a big leak of mine.

I'd like to do a leakfinder session with you but I'm a little bit deterred by the idea that my using PT3 will prevent you from massaging my data (filters et cetera) as you would like.

It's your call, and I am not selling, but I have equal facility with either program.

Also, I am pretty sure that HEM's RCE is 1.0=break even, not 0 = B/E.

The only time I have ever seen an RCE of 0 was when I had a filter on that was 100% losing hands.

Yeah, I know you don't need to sell

I'll probably give it a whirl sometime in the next few months. I am so bad I'm sure you can find tons of stuff to help me.


As mentioned above, I originally thought it would be 1.0 is b/e but the definition quoted from the HEM site jives with a 0 = B/E. Also, some of the discussion on the linked to HEM forums implied that 0 = b/e. Damn confusing. Presumably with HEM we can't go and look at the SQL code that defines the stat? That could eliminate all doubt.

Well, using my 1.0 = b/e RCE stat that Kraada (PT3 support) more or less wrote for me, my RCE over the last 100k hands is 1.7 and over the last week when I nitted up super tight it was 2.2 so perhaps my river play is not so atrocious.

I still make obvious (even to me) mistakes on the river but perhaps these are not my biggest leaks.

This thread made me look up RCE in my DB.

Holy **** 2.5!!!!!!

Edit to add: I'm going to go ahead and believe that its because I'm good at soul reads, and not because I'm too nitty.

Heh. My RCE at 10NL is 2.95. Guess I suck at poker.

EDIT: Of course, I play almost exclusively deep tables. I'm not sure if that makes a significant difference in expected or "good" RCE, but I'm sure 3 is ridiculous anyway.

bump, i found this thread in google.

HM3 has my RCE at 1.1 over 100k hands in heads up plo.

when I filter "RiverActionCheckCall=true" I have 0.92.

Am I leaking like titanic when check/calling river?

edit: didn't notice this was nolimit holdem forums.