In a 5 point match, I'm 3-2 down
No doubling
Opponent is clearly going to win a gammon (so winning the match 5-2)
Towards the end my moves don't matter and should all have equity value of -2.00
however they are all showing -2.988

does normalized equity mean something different here (as its not just the game but the match)?

Short answer is that it's accounting for the fact that, at this score, winning a gammon is more valuable than usual for your opponent.

If the score was 0-0, and your opponent had a play that won 10% more gammons but lost the game 6% more compared to some other play, then it would be wrong to make that play. But at this score, it would be correct. So the bot needs to adjust the equities to account for differences like this. If winning a single game is +1, then winning a gammon must be more than +2 since that extra point it gets from winning a gammon is worth more than usual.

Google "backgammon match equities" if you want to learn how to quantify this stuff.

thanks
i did find something in another forum
they described NormEquity = -1 + 2*(mwc-mwcLose)/(mwcWin-mcwLose)

mwc being match winning chance (but I don't know if this is probability of a winning a single game (and/or adjusted for g/bg) ..... in my case it was 0 regardless
mwcLose,Win were from MET Win put the games-to-win 2,2 Lose put games to win (1,3) [they assumed single point game]
taking the xG tables used i got -1 + 2*(0 - 24.92%)/(50%-24.92%) = -2.988

but I have to guess that my mwc value of 0 had to include my opponents gammon value [ie mwc is matching winning chances] ... else (jf no gammon) i would imagine my mwc = -24.92% (since i would be ending up lose 1 point and being 1-3 down with certainty and so could derive the value from the MET)
does that sound right?

Not sure about all that. I tend to think of these equity numbers more abstractly.

You don't want to get too bogged down in the weeds here. Use the equity #s as a guide to what the right play is and by how much.

that's a fair point (and good advice). I need work on not going down mathematical rabbit-holes as much as working on the game.

Something interesting building on your original post — it’s possible you might have missed an opportunity to double while behind in this game. I’m not saying yo did, but just that it’s theoretically possible.

Consider a hypothetical position where your opponent wins gammon 78% of the time, wins single 2%, you win single 15% and you win gammon 5%. That certainly is the type of position you might have had at some point. Consider your match winning probability if you don’t double (your opponent should not since you’d have an easy drop and would increase your winning chance by doing so). If you lose gammon, your chances obviously are zero. If you lose single, you have about a 25% chance of winning from 4-2 down. Winning single evens the match — 50% chance. Winning gammon put you ahead 4-3 which gives you a 70% chance. Doing the math: .78 x 0 + .02 x 25 + .15 x 50 + .05 x 70 = 11.5% match winning chances.

Now consider the same position if you double. Losing either gammon or single is the same - zero chance. Winning single now puts you ahead 4-3 giving you 70% chances and winning gammon now wins the match. Suppressing the zeros gives your winning chances as
5% + .15 x 70 = 15.5%. This is 4% higher than you’d have by not doubling, so your double is correct. (Opponent has an obvious take since a drop makes it 3-3, giving you 50% MWC).

Seems paradoxical that you can correctly double when you are a 4:1 underdog, but it just shows that you should not just automatically assume the cube is out of play, especially in match play.

stremba's concept does come up in practice. To demonstrate, here's a plausible position:



Note this is only a double because hitting will lose Black's market (White will have a pass). I see a lot of players go wrong here. They'll see that they're going to get gammoned when they miss, so they double with any shot. But if hitting doesn't lose their market, then they should hit first, make some improvement so they're threatening to lose their market, and then double. This protects against the sequence where they hit, roll badly, and then lose a single game.

thanks for the follow ups. I've just start playing more > 1 point matches and find the doubling cube to be very difficult ....i'm using a lot of hints to get a sense of what the board looks/feels like for normal doubling consideration.
Some of the end game stuff is very interesting though ... where you get a better grasp on the probabilities ..... produces very non intuitive plays