The Phi-T mousegame works like a really simple model for stock markets. Either stock prices rise or fall within a day. The program based on Phi-T NeuroBayes tries to predict these changes. It wins or loses one point depending on whether the predicition was right or wrong.
The upper picture shows the market performance, the lower the Phi-T-strategy. Both start with 100 points. The overall simulation time is equal to two years (about 500 workingdays). As you can see, the Phi-T portfolio develops better than the market in most cases. This is just a simulation, but...
Now here is the best: the market ist not generated by a random-number-generator but by you! You decide how the market evolves:
Press the
[Random walk simulation] button, to get the charts for pure random numbers.
Repeat this several times and you will see that the Phi-T strategy has a 50% chance of winning, just as you would expect.
left mouse button: prices rise
right mouse button: prices fall
The program will try to predict your decision.
As you will see, it is very difficult to move the market in a way which prevents the Phi-T strategy from being successful.
,,You are cheating, aren't you...?''
To show you that the program is not cheating, you can switch from
[Hide predictions] to
[Show predictions]. Then the prediction for your next click will be shown on the right. But now of course you can cheat...
,,That's impossible...?''
You will see this is no whitchcraft if you take real random numbers to generate the market, like those from a lottery which are impossible to predict. (To show you, that this is no whitchcraft)
,,Ok, how do you do it?''
NeuroBayes is able to find even smallest correlations and learns from them.
The program looks for patterns in the sequence of your mouse clicks and derives a prediction out of that. If you know that you can beat the strategy by changing your tactic as often as possible. But that is very difficult.
,,You can do even better?''
Yes, we know before each click how sure we are, that our prediction is right.
If Phi-T is not playing with a constant but with a dynamically adjusted stake, the winning potential gets even higher. Try it by switching from
[Simple Strategy] to
[Dynamic Strategy].
Or even to [Exponential Strategy]. Here the stake is adjusted proportional to the already achieved portfolio value. Now we're really flying...
,,Are you playing against us, or with us?''
This little game shows why even monkeys can win against professional brokers at stock markets. The best they can do is to bet totally randomly.
Then the chances are 50%:50%. If you try to be extremely clever, you will lose in most cases. Why? Because the guys on the other side are even more clever...