前注：没有时间翻译成中文，请大家自己使用翻译软件阅读。
- Z0 B# k. f+ |& O! \' e
3 [% K. l8 I" MExplanation of fair odds% F! u. t2 x, q) h2 ?$ r
" r Y1 t+ n8 B/ J; M
There are two kinds of odds: line (market) odds, offered by a bookmaker, and fair odds, presented in this page. Fair odds equalize opportunities of a bookmaker to have a profit, and a bettor to win. In decimal notation, they are equal to the inverse value of probability. Synonyms are true odds, refined odds. Line odds are based on the betting volume, and reflect subjective expectations of thousands of bettors worldwide. Bookmakers should set the line odds less than corresponding fair odds as long as they want to have a profit in the long run. Lucky bettor finds those line odds that are greater than fair odds. The latter may have place e.g. due to the bookie's mistake.
7 W) k, `' j8 C# \' w; ~ E$ t
Warning for serious bettors
) a0 O, ^; p/ | ^- n/ ^/ x
9 ]2 d5 j2 i& |Serious bettors must be aware that they are doomed to waste 10-20% of their bankroll in the long run if bookmaker's line odds are less than fair odds. The bookie's mistake can make the line odds greater than fair odds. Therefore only pairs line > fair can be considered worth for betting.
" B4 R( N: W1 n9 y3 o
$ V% \2 Z3 D$ M: [7 z' }. CMaking betting decision, L) y1 r0 i2 W
2 S K. b) e. |# q* L; J" W) Y
Making betting decision is a complex task. What pair of line/fair odds is preferable for effective betting? What bet for example is better: A=1.3/1.2 or B=1.7/1.5 ? The answer is none of them but the double parlay AB=2.2/1.8 ! Of course any single like C=2.2/1.7 or C=2.3/1.8 will supersede the latter.
% F$ i8 k4 ~- f" Y! E8 e4 k2 q
% H! u/ J: B1 J" K. Q+ K* r
statistics2 {& p/ {6 F( s
1 B, t) ?4 _, m
First table represents a summary of performance/predictability. Exact meaning of its columns:
) E) B+ U& A7 Z7 W* a# d7 L% P: t
( C0 @- F& H( I; F6 a3 g
| Actual number of games [td]Number of home wins (n1), draws (nX), away wins (n2), and total games (N=n1+nX+n2) within analyzed period |
| Relative rate, % [td]Ratios 100*n1/N, 100*nX/N, 100*n2/N, and their sum = 100% |
| Sum of computed probs [td]Sum of computed probabilities for each game to be: home win (p1), draw (pX), away win (p2), and their sum pN = p1+pX+p2 = N |
| Relative rate, % [td]Ratios 100*p1/pN, 100*pX/pN, 100*p2/pN, and their sum = 100% |
| Successful computed probs [td]Sum of successfully computed probabilities for home wins (s1), draws (sX), away wins, and their sum sN = s1+sX+s2 |
| Success rate, % [td]Ratios 100*s1/p1, 100*sX/pX, 100*s2/p2, and their sum (PP) |
, f& I4 R, _9 IThe last sum of success rates PP = 100*s1/p1+100*sX/pX+100*s2/p2 is considered an integral performance/predictability index. It is highlighted by bold font.
; E0 n( E' c- S+ h& s3 e) j- t' j$ Y8 C& ?
Second table shows how many wins or wins+draws (number and %) had actually taken place at various levels of computed probabilities for home (away) wins for those events. For example, the line
3 l" {6 ^4 C& U* l; a1 h( I/ l
* G% N1 Z: M6 E$ u' B7 \Computed Success Wins+ Success
7 _' X! u5 a y9 C7 t6 d5 v! c
probabilities, Games Wins rate, draws rate,
: m8 s* ~' l' }+ K1 ]6 a, a6 {% a3 H, V %, >= % %
7 K9 k" Q% v: o1 p& H/ ^* ^ 85 305 277 90,8 300 98,4
/ e6 Q& [# o; N) H4 C% O2 v $ G! ]0 B. O; I; f; }$ \0 \- o5 z
indicates that in 305 games computed probabilities for home wins (
1) were equal or greater than 85%, and there were 277 successfully predicted home wins (
1), and 300 successfully predicted home wins+draws (
1X).
