The Odds strategy is a mathematical procedure from the decision theory to select it made possible with large probability an optimal "opportunity "from a consequence of events. The procedure developed of the Belgian mathematician F. Thomas Bruss.
The strategy can be used then, if a temporal succession of events is present, by which some as "opportunities "applies, and before occurring an opportunity it does not admit is whether still another better opportunity follows later. A popular example is the situation of a used-car dealer, who does not know when being present a bid whether a further prospective buyer makes a better offer later.
In order to be able to use the Odds strategy, the reality must be mathematically modelled. In addition a consequence of n events is accepted, e.g. each event could be a bid. The events are durchnummeriert with the index k from 1 to n: E1, E2," Ek," EN
Each event Ek is with a certain probability pk one "opportunity ".
If pk the probability for it is that Ek is the looked for opportunity, then is q_k = 1-p_k the probability for the fact that it is not it.
The strategy of the quotient has their name r_k = \ frac {p_k} {q_k}, which is called English Odds.
The strategy functions simply by the fact that starting from a certain index s, which "stop index "the first opportunity is noticed. In the example of the used-car dealer that that the offers of the first s-1 does not accept customers and notices starting from the event it the first opportunity, is called thus that customer sold, who s-ten as the first after that a better offer than all its predecessors makes.
The stop index s is determined, by odds are backwards noted: rn, rn-1, rn-2 etc. thereby are summed them, until the sum 1 is reached. That k, with which this sum is reached, forms the stop index.
This strategy is mathematically provably the best under all the strategies, which must select an optimal opportunity on the same assumption.
The probability for the fact that the best opportunity is used computes itself from the sum R-S rk and the probability of the QA for the fact that under the which are applicable events no opportunity is.
Then the probability of success W = R_s is \ cdot Q_s.
| k | p_k | q_k | r_k | \ sum_ {i=n} ^k r_i |
|---|---|---|---|---|
| 16 | 0,0625 | 0,9375 | 0,0667 | 0,0667 |
| 15 | 0,0667 | 0,9333 | 0,0714 | 0,1381 |
| 14 | 0,0714 | 0,9286 | 0,0769 | 0,2150 |
| 13 | 0,0769 | 0,9231 | 0,0833 | 0,2984 |
| 12 | 0,0833 | 0,9167 | 0,0909 | 0,3893 |
| 11 | 0,0909 | 0,9091 | 0,1000 | 0,4893 |
| 10 | 0,1000 | 0,9000 | 0,1111 | 0,6004 |
| 9 | 0,1111 | 0,8889 | 0,1250 | 0,7254 |
| 8 | 0,1250 | 0,8750 | 0,1429 | 0,8682 |
| 7 | 0,1429 | 0,8571 | 0,1667 | 1,0349 |
Assumed, the used car salesman knows that in one month average 16 customers are interested in a car, and it would like to naturally sell to that customer, who offers the highest price. An event is for the used-car dealer thus one "opportunity "if it is better than all previous. To the first offer applies with security, therefore is p_1 = 1. For the second offer p_2 = \ is frac {1} {2} etc. general applies p_k = \ frac {for 1} {k}.
From this q_k = \ frac {k-1 follows} {k} and r_k = \ frac {p_k} {q_k} = \ frac {1} {k-1}.
Since the used-car dealer has on the average 16 customers per month, is n = 16. The accompanying table shows that the stop index is 7, because with k=7 the sum 1 is reached.
The used-car dealer must wait thus up to the sieved offer, and then first to assume, that is better than all previous.
The probability of success is W=R_s \ times Q_s = 1.0349 \ times 0.3750 = 0.3881, thus approx. 39 %. In other words: The used-car dealer sells the car in 39 % of all cases at the best price.
We found here 6 related websites.
Index | Privacy | Terms Of Use | Sitemap | Feedback