An Indicator to Play with

Here is an idea I want you to test. Let us call an event that depicts the change in both inside Bid and inside Ask levels. Let us call this event a “unique price change”. Let us assume that those changes are the result from a “Random Walk” type of a process. It is a well-known fact that the binderies for such a random walk (its mean deviation) could be calculated as squareroot(2N/pi) where N is number of unique steps (in our case it is the number of unique price changes described above). If one can draw those binderies on the chart in real time it seems that every time that the close price of a bar (it does not matter what the bar length is, I tested it briefly on 5 min bars) ends up outside of those binderies it has a strong tendency to reverse it self. I have briefly tested the idea on ES contract. I also included in this post a code for eSignal (it seems to me that many people on ET still use it). I would like you to test the idea in real time and discuss it. May be we will find the way to improve it and use it efficiently for the intraday trading. Looking forward to your responses.

P.S. to test the idea on other securities you need to change the tick value from 0.25 to the tick value of the security you are trying to test. Also, the lines will not appear on the history bars because bid/ask numbers could only be retrieved in real time. Play with it.

 

Interesting enough, I placed 5 trades today so far based on this indicator; all of them worked out well. It could be a coincidence, of course. But it is awfully intriguing!

 

If you measure its position (current price) you will change its energy (price change) or vice versa, but you can never know both at the same time. By giving the indicator/code to all the flying monkies out there it will cause them to jump on board and it will no longer work.

But thanks for the thought

 

I do not subscribe to this theory. If you take all of the ET members with all of their trading capital it will be still a drop in the bucket! With this philosophy there wouldn't be open source developments of any kind. Look at Google, Sun etc. they seem to do well with the sharing concept! I believe in collaboration; call me naive.

P.S. BTW, I made 7 successful trades today on ES with the average profit of 0.75 points per trade. I would not call it the earth shuttering result, but it is very encouraging.

 

Hate to bust your balls but I have some free time ----

Google etc are entrenched monopolies --- they share because it benefits them to do so. My description of the Hiesenberg Uncertainty principal (1927) as it applies to sharing whatever with others will apply once some critical mass of use is achieved.

On your indicator --- the Gaussian distributive process is based on a normally distributed process that is mean reverting --- which will lead to some very unfortunate outcomes anywere from 2 to 15% of the time depending on market conditions ---- power laws are a better description of market behaviour. Although you could have a good string of wins for a length of time. I would advise you explore that route as well.

My 2 cents ---

Regards,

 

Appreciate your response. My thoughts were based on the assumption that the actual price distribution in the short time frame is very “skinny” power law type of the distribution. That is why using Gaussian binderies I was hoping to find the reversion not to the price level but to the coincided peaks of these two distributions.
I have now completed the history run on ES for 3 month (tick-by-tick data files). It shows definite dependency there.

 

I don't have eSignal. Does the code run (trigger) one per bar, or once per tick (last price change) or every change in the bid or ask price?

If the code runs once per tick is it basically doing the following:

1) at the beginning of each new bar, numPriceChanges gets reset.

2) get the current inside bid and ask price (the bid and ask that exists at the time this tick occured).

3) if currBid (price) and currAsk (price) change (on the tick being processed vs the previous tick processed) then increment numPriceChanges.

4) calc theorValue, upperLine, bottomLine.

5) plot upperLine, bottomLine.


thanks,
ct

 

Yes, that is correct.

 

WTF is a "bindery" in this context? Do you mean "boundary?"

Itself, BTW, is one word, not two.

You claim several advanced degrees. What university awards such degrees to an illiterate? Surely not one stateside.

 

1) is it supposed to look something like this?

2) are these correct entries on the chart below? if going for 4-5 es tics at a time with a small stop, this random day i used amongst my tic files looks good at first glance

[/img]