"If you only bet overlays, you can only win in the long run." You�ve read these pronouncements. Strictly speaking, it�s a true statement. The rub, of course, is deciding who is an overlay before the race. It�s like saying, "If you only pick good mushrooms, you�ll never die of food poisoning."  Well, sometimes the bad mushrooms look like good mushrooms -- kind of like overlays at the racetrack.

I�m a definite proponent of value betting, but my experience with betting overlays is that it�s streaky. I�ll go through weeks of good profitable betting, and then blam!, I�ll suddenly enter a losing streak where nothing I do is right. Now, losing streaks are a fundamental part of the life of a handicapper, as I�m sure every one of you can attest. After years of bitter experience, I learned to cut back during a losing streak, even though it didn�t strictly make sense within my overlay ideology.

But what causes these overlay losing streaks? I realize that there is a statistical variation inherent in any series of wins and losses at a given hit percentage. But there seem to be multiple reasons for overlay losing streaks beyond statistical variability. One of the most important is betting action itself. The very thing that defines an overlay � a horse�s odds � is also a strong predictor of how often that horse will win.

Let�s step back a moment. Sometimes a thing won�t make sense until you find an important distinction that throws light on the subject. For instance, an old study of maiden winners showed them winning 14% of their next starts at an ROI of .72. That�s pretty close to the numbers you�ll get from random betting. However, when the distinction of the margin of their victory was made, it was found that horses whose first victory was by three or more lengths won 20% of their next starts at an ROI of .99. That�s a meaningful distinction.

Similarly, when you look at the effect of betting action on value horses, an interesting distinction comes to light: in some situations, the classic overlay is profitable, but in others, one would get slaughtered betting overlays blindly. Why? Because, in these situations, virtually all the winners were horses with betting action, horses being bet down below their fair odds. In some situations, unless a horse is christened with betting action, it has very little chance of winning. And if you�re betting classic overlays in that situation, it�s like spitting into a headwind � the outcome won�t be pretty.

These tote-action horses look like classic underlays, but they�re not, because they often actually show a profit. A group of "underlays" that generate a profit are by definition not true underlays. One solution to this dilemma would be to modify your betting lines to incorporate tote action. That way, tote-action horses would be assigned higher probabilities just because they�re being bet. But that�s extremely impractical unless you have a computer system hooked up in real time to the track�s parimutuel information. More practically, we can recognize the situations where tote-action horses trump overlays.

For instance, using my computer program at Charles Town over the last five days (I�m usually more interested in adequate short term samples then long-term samples, but that�s another article.), there were 15 tote-action horses on the top of the betting line. Five of those won for a 33% hit rate and a 1.13 ROI. In contrast, there were also 15 classic overlays on top of the betting line. Of these, only one won, for a hit rate of 7% and a .26 ROI. If you had gone blindly into this situation betting the classic overlays, you would have gotten stomped. But by recognizing the fact that Charles Town (like many bullrings) is a tote-action track, you could�ve had a much nicer win percentage and a profit to boot.

Looking at the last week for the 45 tracks where data exists, and ignoring the profitability of horses other than the top pick of the betting line, 12 (27%) of them were profitable with classic overlays only, 8 (18%) were profitable with tote-action horses only, 5 (11%) were profitable with both classic overlays and tote-action horses, and 20 (44%) weren�t profitable in either category. If a horseplayer were to blithely walk in to the 62% of the tracks where classic overlays weren�t profitable (at least for the top pick), he�d walk out with less money.