Hope this forum is appropriate for this.
I was playing around with PT4 filters (over 16k hands) to figure out how my river bluffs are working out and I'm kind of confused.
These are my results when I bet river when heads up OTR (at least I think, please check the filters in pic to make sure I'm not misinterpreting):



So winning a lot in red line, pretty good. However when I add the filter that I bet with high card here's what happens.



I would appreciate it if someone could help out with a couple questions I have about this:

1. Is "bet river HU with high card" a good enough approximation for "bluffed river" or can I use a better choice of filters? Kinda surprised to only see 50 hands as I didn't think I was that nitty

2. Do these graphs imply I should simply bluff river even less? Again confused because the first graph is winning at non SD so the only conclusion I can make is that my value bets are getting a ton of folds while my bluffs are getting called over the sample.

3. Is the discrepancy simply because the sample isn't big enough?

Many thanks!

What did you expect? That you're making money in spots where you don't even have a pair otr?

Both graphs are normal.

I'm echoing ZKesic, but I don't see a discrepancy.
When you bet the river, sometimes it's for value, so you do well whether villain calls (your blue line goes up) or he folds (your red line goes up). When you only have "high card" (you're bluffing), you almost never win when called, so you only make "redline" profits, and the blueline is horrible.

There's no doubt that you make errors on the river, because everyone does (no one balances their ranges perfectly), but I don't see anything abnormal in these graphs. Betting the river in general is profitable. Some of the profit comes from value hands, some comes from bluffs. The bluffs make most/all of their profit from getting folds. You can't expect the bluffs to also make money when villain snaps them off!

P.S. I'm not sure if you can do it with PT4, but in HEM you can run filters like "bet river" and get it to make pie-charts to show you what your river betting range is comprised of. That can help you work out if you're over-bluffing (or not bluffing enough).

Is it possible to set a PT filter to "hand strength < TPGK" or something like that. I would presume that you sometimes bet a weak pair as a bluff, and that should bring up your proportion of "non-value bet hands" to a few more than your current 50 out of 391 total bets. Although it might be fine to seldom bluff in your game, especially if you often use small sizes on the river, only bluffing about 13% of the time would indeed be nitty.

FWIW, here's what HEM gives me if I run the line analysis on myself for the filter "bet river" on a small sample of recent microstakes hands.



As you can see, I was betting with TPGK or better 68% of the time. I only had total air (high card) 21% of the time, but I also bet some hands that were worse than TPGK as bluffs (maybe there was some super thin value with TPwK). My 68:32 ratio might actually be a little bit "too bluffy" given that my river bet-size is usually less than pot, but I suppose my range is reasonably balanced. If your range was truly weighted in a 341:50 ratio, it's surprising your bets ever get called, as you'd be extremely value-heavy, and observant villains would stop paying you off.

FWIW, I ran the "bet river" filter on my database, and then ran it again after setting hand value to "top pair, weak kicker, or worse" to see if anything odd happened. I got these graphs for a similar sample size to OP's.

Bet River:


[x] All lines going up.

Bet River with TP weak kicker or worse:



[x] Broke even with bluffs.

I must be some kind of GTO bot.

EDIT: If I change the filter for handstrength to strictly "high card", the green line drops to -140bb, so some of my bets with weak pairs must have got called by worse.

Tyvm, very insightful! That last graph is sexy

That's simply not even remotely close to a relevant sample. All you are looking at is noise. You can brute force your way into relevant exploitative play when dealing with situations that come up very frequently, mainly preflop, but river situations come up so rarely and are so varied in their nature that looking at river samples is borderline useless.

For low sample spots it's much better to just rely on theory.