There are two extremes in "overlays" and "underlays" that you need to be aware of. One is the true low-odds  overlay, the other is the high-odds underlay.

The way to understand either one (as well as the more conventional mid-range overlay) is to first, of course, understand your own handicapping, and second, understand how the crowd handicaps.

All of us love the \$21 horse. We'll always remember the time we saw the hidden potential of a horse that the crowd placed at 20/1 and we were right - and they were wrong. For probably the majority of handicappers, when  they truly handicap a \$21 horse (as opposed to taking a longshot flyer) the money is nice but secondary.

It is validation of everything we do. The crowd saw this horse as a 20/1 shot, which means they gave it only a 4.76 per cent chance of winning. Conversely, that's a 95.24 per cent probability of losing. We saw beyond that, understood the race better, and gave it, say, a 3/1 (25 per cent) chance of winning, and took the bet.

So does this mean that if you "like a horse" and it's on the board at 20/1, you take the roll out of your sock? Only if you want to be selling popcorn in the autumn.

The only way to evaluate what the crowd is offering for odds is to make an odds line yourself. "Liking" a horse has no place in "value betting" unless you know exactly how much you like it. That involves both your own handicapping and skill, and making an odds line from it. But just as importantly, it involves understanding what is shaping the crowd's odds on this race.

The smaller the difference between your odds and the crowd's odds, the less there is to understand. If you are handicapping a routine race that fits your methods, and you make the win contenders 3/1, 4/1, and 5/1, while the crowd reverses those odds, there can be a myriad of reasons, but they are probably ones you can readily see: a rating figure that they buy and you don't, a spell that you see as positive and they see as negative.

You size up the race differently and you can bet your 3/1 horse at their 5/1 offer, or work a strategy on an exotic with routine confidence.

Suppose, however, that this is the \$21 horse. The crowd makes it 20/1 with all the above statistics. If this is a  nine-horse race and you make this horse 15/1 and the crowd is offering 20/1, is that an overlay? No. You are giving this horse a less than random chance of winning, and it is not an overlay at any price.

Now, suppose this horse is your third choice in the race and you make it the 5/1 horse among the three contenders above. Is it an overlay for a major win bet? Some advocates of value betting would say "Yes".

I have heard it argued on a purely statistical basis that if your 5/1 horses win their fair share of races (16.67 per cent), and if your first two choices are standing at fair odds or underlays indicating no "value" available from them, then pound your third choice for the percentage angle.

There's only one little problem here. If I give a horse a 16.67 per cent chance of winning a race and it is my top pick as I understand this race, then chances are it will win something around that percentage of the time.

However, if I give a horse that same percentage chance to win but give two other horses better chances, which the crowd agrees with and is heartily betting them into underlay range, then that third choice of mine is probably not going to win its projected 16.67 per cent share. It might not be a "win" contender at all, but a "money contender" (place chance).

A third-choice, percentage-angle "overlay", for a good handicapper, is usually a rotten win bet. As with random odds, you are betting against your own skill. Just because the public odds on a horse are higher than your odds, does not necessarily make it an overlay, and just because you make a horse an in-the-money contender does not mean it would actually win this race in this lifetime.

So how do you spot an overlay and how do you avoid some of the pitfalls of false overlays? Again, the first element is to understand and trust your own handicapping. Your own handicapping forms the baseline for comparison with the public odds.

If you have followed any of my writings, you have probably noticed I have hardly written a word about "handicapping", per se. That is because I believe that virtually anyone (with a few exceptions) can learn to  handicap. Handicapping is fun. It's work, but it's fun.

The only individuals whom I have seen fail to become  good handicappers are those who refuse to believe that it requires work. Whether you use speed, Pace, class, or all of these, you can learn to be a good handicapper.

Where good handicappers fail to become successful bettors is in the conversion of the results of handicapping into betting strategies and this is NOT money management.

The vast majority of players do not play from a bankroll; they play, as Eric Langjahr says, "per cent-of-pocket", so pontificating about "units", etc. is meaningless. For good handicappers, whether they are large or small bettors,  the trick is to turn the results of their handicapping skills into well designed bets.

These days, there is a nice array of potential bets at most tracks that let you shape bets based on your results. "Contrarian" handicapping takes all this into a new realm which I really enjoy, but for now we'll only worry about conventional handicapping.

Handicappers who think only in terms of win contenders are not only limiting themselves to just one area of potential profit (win overlays), but also exposing themselves to a wider array of pitfalls than if they understand  and play the full variety of overlays available.

Once you have handicapped a race and believe you understand it, in terms of win contenders, money contenders, and non-contenders, then overlays become possible in all pools. Some are harder than others to identify, and we'll talk about some, including my favourite, the Place Overlay, in the future.

In all cases, in addition to the conventional overlay (you say 3/1, they say 7/1, and you know why), it's good to be on the lookout for the two oddities described at the beginning: the low-price overlay and the high-odds underlay.

One makes money; the other loses it (remember the little "edge" equation), without getting too personal here, it's no secret that I had some pretty good alternative speed figures before Andy Beyer's Figures were published. I also had the advantage of working at a track that had one of the most, shall we say, "naive" publics in the nation.

As a result, I became spoiled rotten by the odds I was given. If I didn't get 12/1 on a dead-nuts, sure-thing,  third-place finisher who had my figures in the "960s" in 10 consecutive races and was going to crush a field of winners who had never broken 900, I would pout and pass the race. I'm not kidding.

So when the Beyer Figures became the single-most important factor in setting the public odds and I started  getting 4/1 on horses I made 3/1, I entered a sort of personal "dark ages". At the track I would pass races and pout and curse Andy, or worse, and make dumb bets, trying to wring some odds out of exotics, all because the public now had access to some pretty decent speed figures.

What finally broke this pattern was accepting the existence of low priced overlays. In my own work, I now use what I call "aggressive odds" and print out for myself on paper what 50 per cent and even 25 per cent overlays should be.

Now, if I go to the track and my favourite win contender is down to 3/1 at post time, I don't necessarily tear my computer printout into confetti and toss it in the air. With an aggressive odds line that more closely tracks the  "piling on" factor of the crowd, I may very well have that horse at 9/5, and can just shrug and say, "Okay, fine, 3/1 overlay," and pound it with a total lack of guilt.

And guess what? In total accord with every theory, statistic, and mathematical formula, horses in these low-odd ranges do tend to come in more than horses set at higher odds. I always knew that; I just never capitalised on it because of old-fashioned handicapping ego.

I doubt that anyone else could be that dumb, but if any of this sounds familiar, start paying attention to low-priced overlays.

The flip-side are high-priced underlays. These are not the \$21 horses that we all dearly love to handicap and ID. These are the \$21 phonies. How do you tell the brilliant score from the phoney?

Well, for one thing, because of the percentage range we're working in here, you are not going to be able to do it as often as you will with lower-priced horses. That's a truism that we all sort of know, but hate to really accept.

If you are a "contrarian" handicapper, you are probably playing in this zone a lot and you, no doubt, have your own theories. In this range of odds - say above 8/1 - the increments between steps get much smaller.

Note that the probability of winning for an 8/1 horse is 11.11 per cent. A 10/1 horse is 9.09 per cent. A significant difference, but not huge, given that the odds step is only two. Now look down the list: 15/1 is 6.25 per cent, 20/1 is 4.76 per cent, and 25/1 is 3.85 per cent. The odds begin to skyrocket with very small increments of change in the win probability.

They can also come down in very big chunks based on very little action, especially at small tracks. I once dropped the odds on a dog from 99/1 to 17/1 with a \$2 win bet at the Juarez greyhound track. So, a drop from 30/1 to 15/1 does not necessarily mean the "smart money" has arrived.

NOBODY knows what the true odds are until the race is finished. There is interplay between the odds of all the horses entered and, once the odds leave the primary contenders of the race, there are many factors causing bets to be made other than a burning conviction that these low-interest horses are going to win. The higher the odds, the less exact both the odds and the reasons become.

You can make up complicated scenarios and rules, or you can depend upon one simple test. If you are making a horse 3/1 and the crowd is making it 20/1, and you know the reason why, you may very well have a true \$21 horse.

However, if you run across a horse that you are making a contender at relatively low odds and it is standing at 20/1 and you don't know why, then, whether it is a true overlay or not is purely academic. It may be somebody's overlay but not yours. Next race.

Charles Carroll is one of America's greatest handicappers. His books on speed ratings are among racing's worldwide bestsellers. Check out his work at Desert Sea Publishing at www.desertsea.com.