Start with the trinity: VPIP, PFR, and TA. That is....
Voluntarily Put Money Into the Pot: the fundamental indicator of a player's preflop looseness, and one of the best gauges of a player's hand range. At uNL, you want your opponents' VPIPs to be as large as possible. The more hands they play, the worse they are, and the more money you'll make against them.
Preflop Raise: the fraction of the time that the opponent raises preflop. This number indicates how you will have to adjust to this opponent preflop. First off, if it is incredibly small you need not worry about this opponent raising you off a speculative hand. If the players left to act all have miniscule PFRs, you can limp with extremely speculative hands and expect to see very cheap flops, giving you marvelous implied odds. Take advantage of it. However, if an opponent with a PFR of 2% makes a raise preflop, you can fold most of your speculative hands without worrying. (NOTE: if you're going to be heads-up against a full-stacked opponent with a PFR of under 4% who raises to 4 or 5 BBs, you can very safely call with ANY pocket pair on implied odds alone, because if you hit your set, you'll stack your opponent VERY frequently. The tiny PFR tells you much about their holdings, and most of these opponents will commit a full stack to the pot with a hand that is powerful enough to convince them to raise preflop.)
Total Aggression: the best overall indicator of how aggressive this opponent will be postflop. Preflop play and postflop play are two different channels, and you should not expect a player to be either passive in both or aggressive in both. There are players who are highly aggressive preflop but turn into calling stations postflop, and there are sluggish limpers preflop who go hyper-psycho postflop when they hit a hand. If you see an opponent with an incredibly large total aggression rating, you will probably have to play passively against this opponent if you hit a solid hand, letting them do the betting for you. NOTE WELL: players with a very large total aggression come in two fundamentally different flavors -- the ultra-weaktighties and the maniacs. An ultra-weaktighty is someone whose aggression is high because he folds whenever anybody else bets and he doesn't have the nuts. A maniac is someone whose aggression is high because he bets and raises with any random crap. If you can tell these two apart, you will make MUCH more money. (Hint: look at the kinds of hands that they show down. If they're reaching showdown infrequently and always with a monster, they're probably ultra-weaktight. If they're losing most of their hands at showdown, they're probably maniacal.)
Beyond the trinity, here are the stats on my HUD:
Folds to Continuation Bet: probably the next-most useful statistic in my HUD, and one that I rely on VERY heavily in making decisions. The more often a player folds to c-bets, the more likely I am to (a) raise preflop with them in the hand and (b) bet into them when I miss the flop. I see no reason to c-bet at someone who never folds to c-bets; instead, I'll wait for a real hand and then hammer the opponent. Alternatively, if an opponent folds to c-bets 60% of the time or more, I'll use my position mercilessly, raising against them VERY often and stealing pot after pot after pot on the flop with a sturdy bet.
Attempted to Steal the Blinds: very handy for figuring out whether an opponent is positionally aware. From CO and Button, I treat this number as the person's PFR%, and I adjust my behavior accordingly. Also, if the person's steal percentage is markedly larger than their PFR, I assume the person has more of a clue about how to play the game, and I give them a bit more respect when they go to war. I also figure that I've got much more folding equity against them in general, since savvy players are less likely to be calling stations and more likely to be capable of folding a decent hand.
Folded Big/Small Blind to a Steal: another basic indicator of hand strength. If the person folds to steals all the time, then I know that when they call a raise from the blinds they've actually got a hand. If the person is a habitual blind defender, I can assume they have crap even if they call a sizeable preflop raise. Combined with the Fold to C-Bet percentage, these numbers tell me whether stealing against an opponent will likely be profitable or not.
Went to Showdown Percentage: the larger this number, the more of a calling station the opponent is. There is no "magic number" here that tells me an opponent will be profitable or unprofitable; rather, this number tells me how I should start to adjust to this villain. Against an opponent with a big WSD%, I value bet lighter and bluff much less often. Against an opponent with a very low WSD%, I value bet much less often and bluff much more frequently.
Won Money At Showdown: another very useful statistic, this is a rough measure of an opponent's postflop skill. The larger this number is, the more likely a villain is to have the goods when we reach a showdown. It helps me decide how lightly I can value-bet an opponent, and it helps me decide how lightly I can call an opponent down. It helps me sort good LAGs from bad LAGs, nits from TAGs, and maniacs from smart but aggressive opponents. If an opponent's W$SD is extremely high, I consider them weak tight and bluff frequently. I also shrink the size of my bets, knowing that it won't take much to push villain off his hand. If an opponent's W$SD is extremely low, I value bet with all sorts of crap, expecting my opponent to call me down. I also raise the size of my value bets, fully expecting to get paid off even if I push hard.
Total Hands: an incredibly important number, this tells you how much trust you can put in all the other numbers in your HUD. At about 50, numbers like VPIP and PFR start to be meaningful. At about 10,000/VPIP you can start to rely on total aggression (e.g., 200 hands for a player with a VPIP of 50, but 500 hands for a player with a VPIP of 20). W$SD doesn't really lock up for several thousand hands, so treat it with a grain of salt, but WtSD becomes meaningful noticeably faster.
Beyond these numbers, I also keep up six numbers about ME: VPIP, PFR, total aggression, attempted to steal, folded to c-bets, and hands. PAHUD keeps track of these numbers separately table-by-table, so you get an at-a-glance look at how you've been playing at a particular table, which will help you identify your table image among a particular set of opponents. When my VPIP is large, I know I'm looking donkish. When my PFR is large, I know I'm going to get less respect from my preflop raises. When my TA is large, I know I'm a 2+2er. When my steal attempts get too frequent, I start to lose folding equity and need to play it a bit more cautiously. If I'm folding too often to c-bets, some opponents may start making moves on me on the flop. Staying aware of these numbers keeps me adjusting properly to the ever-evolving table conditions.
The last thing I always have on my HUD is the showdown information. By having the board and the hands for every opponent included in the HUD, I don't have to constantly open up the hand history just to see what my opponents had. It helps me to judge how they are playing, and that's vitally important information for any winning player.
One more point: I have basically every piece of information that PAHUD can display included in the pop-up screens. That way, if someone makes a strong move at me, I have instant access to everything I could need to know. How often do they win when they check-raise the turn? How often do they cold-call preflop? How often do they raise continuation bets? How often do they win when they see a flop? How often do they check-raise the river? You never know when one of those random factoids could wind up saving you a stack, and since it costs you nothing to have that information included (no clutter, since it's only in the popup), you might as well give yourself access to it.