Skill level is very hard to pinpoint in a game where luck is an important factor and players play different games. How could you compare a 50k NLHE sng player to someone who plays 200/400 Pot limit omaha hi/lo cash? They are two completely different games thus its hard to compare. If anything, you could consider an Elo strategy in just cash games, for example, everyone starts at 1500 as an elo. If you win some amount of BB versus a player, your elo goes up, and vice versa. It could be weighted so that if you win BBs off someone who has a higher elo you win more points and vice versa. Just a thought, would probably be very very hard to calculate and I Iiterally just thought of this five minutes ago.
Right!
For example, in the last 3 days, I have played 9 MTTs. I won 4 of them, 2nd in 1, and cashed another. The tournies i won were 2 NLHE, 1 PLO, and 1 Omaha.
How does one go about rating my skill (or lack thereof) relative to a NL cash player? Can it even be done?
Yes, I think any rating system can only apply to a single game, although even what constitutes a single game can get a little unclear. Still, no one would expect a chess rating to translate across to a given go rating, and in the same way, I donât think anyone would expect a rating for no limit holdem cash games to translate across to 7 card stud. The first two are both board games, and the latter two are both card games, and in both cases there is probably some similarity in the skill set required for an individual to become strong at both, but in both cases it is very possible for someone to be very skilled in one, without even knowing how to play the other.
1 Like
Actual value vs expected value is probably the best metric assuming it could be calculated. You have to make assumptions about your opponent to do that though. If you assume theyâre playing GTO, the metric gets skewed the further away they are from that. It ends up favoring people playing against the âworstâ players, they same way BB/100 does.
That raises an interesting point though. You could argue the best poker player is the one that selects the games where they have the biggest edge, and their skill at the table is secondary.
In the end, absolute number of chips is probably as good a measure of skill as any other.
2 Likes
i agree. Moreover, I would say that BB/100, at specific stakes and number of seats is as good an indication as one could hope to find⌠in ring games anyway.
Thatâs little use to me as a tournament player, but you werenât trying to find a metric that would meet my needs. Sorry if I tried to expand the thread beyond its original intent. (No, that wasnât sarcasm)
This is a topic I find interesting, and one that has been kicked around here for a number of years.
2 Likes
Iâm not that familiar with chess, but my understanding is that ELO is relative - that is, you gain more points by beating stronger opponents, and lose more when defeated by opponents with lower ratings. I think any good rating system has to factor in opponent strength somehow.
A lot of video games face similar problems when matchmaking - there may be something in that realm that could be applied to poker.
Yes, you are correct about that aspect of ELO ratings (and other, earlier rating systems used in chess also). For example, if you had an 800 point advantage and won, youâd gain almost nothing, while losing that game would give you a maximal loss in rating points (and be a giant gain for the lower rated player).
1 Like
Total chips are certainly the goal. But I think the main limitation with comparing total banked chips is that it is more a measure of career achievement than of current strength. By that measure, someone who has been playing longer has a propensity to be âratedâ higher. I think with ELO, for example, games from the past gradually diminish in the extent of their contribution to current rating, as they are less indicative of current strength. (I canât recall how that works mechanically, though).
2 Likes
Hi Yorunoame, Interesting post. Can I ask though how you tracked your performance here on Replay to calculate BB/100. (I assume you didnât manually record each of your hands and I canât see a way to (easily) download hand history here on replay)
To further this convesation, I myself am interested in creating a âDid I play wellâ KPI. I have recently started playing money sites after joining replay over a year ago. but I was immediately struck by how different it was.
I wanted therefore a way to strip out the luck effect , and actually have a way of measuring âhow did i playâ. Because of course in poker there are some hands you lose, but you would play them exactly the same (so even though i lost i did not play badly here), while there are some hands you win, when you donât deserve to. (getting your magic river card when you should have folded, so here you did play badly)
I launched a bunch of test runs for the âComparing Simple Strategiesâ thread, and before beginning each run, I would just jot down my current chip level, and hands played. You can find the number of hands played by going into âStatisticsâ.
Iâd also use the statistics section to figure out when I had played 1,000 hands in the run (I usually over-ran it slightly). At the end of the run, Iâd then go into the âBankâ section (higher up in the same screen shot), mainly to be able to see how many $2,500 bonuses Iâd received during the run so I could subtract those out. Obviously, on 50k/100k, or something on the higher end, those made no real difference, but on the lower stakes run like 1/2 or 2/4, subtracting those out was vital to get any kind of meaningful number.
Iâd note that there is still quite a bit of result volatility in 1,000 hands⌠even in a case like this where my advantage was very large. Youâll need probably 10,000 hands or more before things begin to smooth out, especially in an online cash setting where players are so much stronger, and youâll be doing very well even at the lower stakes to win 10 to 20 big blinds per 100 hands (and 3 to 5 at higher stakes would be crushing those games).
Edit: oh, in âBankâ, youâll want to go into âsee allâ to be able to find all of your bonus payments to pull those out.
1 Like
The idea of ranking poker players is an interesting one, and it gets at the heart of how difficult it is to rank players in poker or any other game/sport.
With the upcoming NFL draft, we see players having pro days (in previous years there was also the combine), where they demonstrate âmeasurableâ skills, such as 40 times or bench press reps. But how much do these measurables tell us compared to what we see on game tape? The game tape tells us how players play against different levels of competition and across different certain situations.
Similarly, when it comes to poker skill level, I prefer the âgame tapeâ over measurables such as bankroll. I assess a playerâs skill level from observing their game session-after-session, focusing things such as their post-flop edge and how they adjust to different players and different situations. The best players can force the table to adjust to their play. This tilts my ranking more towards something that is qualitative than quantitative, which cannot be assigned a numerical rank such as in chess.
In any case, how would knowing a playerâs rank help me against them?
1 Like
I think less skilled players make way more frequency-type mistakes and these are pretty easy to exploit. Overfolding, calling too much, bluffing not enough or too much, 3 bet frequencies, and so on.
Knowing they are less skilled, itâs pretty easy to find and exploit their mistakes.
Better players are harder to exploit because they generally play a more balanced game. However, since these players will adjust to my game, itâs usually not that hard to force them to make adjustments that benefit me.
Usually, this involves showing one play style until they adjust, then switching gears in order to capitalize on their adjustments. This is easier said than done, but it can work.
I think that understanding how your opponents approach the game is one of the key factos of successful pokering. Ranking is just one more piece of the puzzle.
1 Like
I started the thread mostly in response to an on-table discussion last week, featuring the frequently heard gripe that poker is all about luck, and there is really very little skill in the game. Iâd be likely to make that complaint about 5 card draw (though even there perhaps that just reflects my own lack of skill in that game), but feel it is so diametrically at odds with the truth when it comes to Holdem. So I wasnât really thinking (while writing the initial post) that there was any real practical advantage to be gained from picking out strong players; I was just trying to say that I think poker is a game there is immense potential for skill, and in fact I think it looms large in that regard when compared to most of the other games in the world (Iâd rank go as #1).
That said, I do think you want to be able to pick out the good players from the fish for quite a few reasons:
- table selection: if you want to build bank roll, its important to have at least one player at the table that you think will bleed chips to you
- seat selection: you typically want to sit to the left of really strong players, or as far away from them as you can get (to minimize how frequently youâre involved in hands against them)
- pre-flop ranges: stronger players typically construct their ranges very differently from typical Replay players; a raise or 3 bet from most players here represents a fairly tight, small range of hands, and 3 betting or 4 betting over that range is not a casual activity, but against a stronger player, youâll often be looking at much wider ranges in those spots
But of course your basic point is valid: rather than overly generic terms like âstrongâ and âweakâ, youâre better off paying attention to individual frequencies and patterns, and just deciding that a player is âstrongâ still leaves a lot of wiggle room for players with very different styles.
1 Like
I understand the interest in rank, but I would further argue that part of the skill of poker involves deducing your opponentâs rank on your own. If you sit down at a table in a casino, players arenât wearing badges that tell you what skill-level they are nor are there HUDs that provide real-time info on their frequencies. All of this information is gained through observation.
If you sit down at a table in a casino, players arenât wearing badges that tell you what skill-level they are nor are there HUDs that provide real-time info on their frequencies.
That would be cute if they did, though. I might need to get a T-shirt with my VPIP and other frequencies printed on it, and wear that on my next trip to Vegas. I could even add juicy tidbits, like âfolds to donk betsâ, âtight UG, but gets wider closer to buttonâ, âAggrotardâ, etc. LOL
2 Likes
The funny thing is that nobody would believe that your shirt had your actual stats.
1 Like
I donât think there is a T-Shirt small enough to fit my VPIP.
2 Likes
For what itâs worth Iâve been playing some real money poker, and my win rate over 5000 hands of 10nl was 25bb/100 and then my winrate over the next 5000 hands was -15/100, which says something about the validity of winrate unless you have 100,000 hand sample. A few all ins can go a long way over 10,000.
Were you playing live or online? By 10nl, do you mean $5/$10??
But in general, when you have a giant edge, it takes more to disrupt equilibrium, while if you have a 3bb/100 edge⌠letâs imagine a picture of a 3bb win rate canoe verses a 200bb win rate oil tanker; what seems like large waves to the canoe are calm seas for the tanker, and so with a really high win rate, normal volatility starts seeming like a placid lake.
In stiff cash games, where edges are smaller, you often hear of stretches running bad that can run 20k or 30k hands. Conversely, here, on 1/2, where I can go all in pre-flop and typically get called by at least half the table⌠yeah, thatâs a little different.
1 Like
This is online .5/.10 6max. After the first 5000 it seemed easy. Suddenly Iâm not so sure what to do after missing the flop and missing draws over and over. It probably doesnât help that in that span Iâve gotten AA/KK all in pre 9 times against dominated hands (70+% equity each time) and lost 6.
During the winning span it would have been easy to conclude that the sample was sufficiently large to infer that my winrate was at least 10/100 but now its unclear if Iâm a winning player at all. So itâs probably less like the oil tanker, more like the kayak with a slow leak in the bottom. But it is interesting to see how much of what appears to be a decent sample can be variance.
3 Likes