Protecting Multicharts Code, best ways?

[i am nobody]

I use Rightedge (a third party program). I'm investigating using Multicharts as well.

Quite simply, I find it insane to believe that Rightedge or Multicharts would wager all potential future sales of their product against whatever results my algorithms would provide them by stealing it. They are as trustworthy products as you'll find anywhere.

They will not steal but COPY without you noticing it. So they take no risk. That's the advantage of electronic stealing, it can be done without somebody even noticing it. Because the don't actually steal, they COPY. You will never miss anything.
And if they ever find something, how will you know they have or not? They will not post on internet that they copied interesting code.
Paranoia is not good, but being naive isn't either. Healthy paranoia is the best attitude.

 

[userque]

Then why does nearly everyone in the biz conceal their signals? As strongly as they protect their code?

Maybe they're like you, believing in something without knowing why--because everyone believes it or does it. Maybe they are using simple systems? Many folks reveal their signals, for a price. You don't have to conceal your signals, just don't tell anyone? I'm not really sure of what your saying here.

Consider this. If I gave you 'signals' that corresponded perfectly to swing tops/bottoms of a market. If you could reverse engineer them, you'd have a system that traded exactly at the tops/bottoms of the market.

Those perfect signals are there for anyone to see and reverse engineer. Why has no one done it?

Reverse engineering code is far from looking at signals and determining how they were generated (SVM, NN, kNN, Lottery numbers, Etc.)

Code can be reverse engineered because it has to be 'run' on some known platform. That platform can be emulated. (Decompilers, Disassemblers, Etc.)

Signals are not like that. They are more like encrypted data. (There is strong encryption and weak encryption)

If I tell you my signals are based on moving average cross-overs; sure, you can decode them.

But if I tell you nothing about the origin of the signals AND if they are based on custom indicators (historical pattern matching, for example), etc. or some very complex system (machine learning, for example); for all practical purposes--it can't be decoded.

 

[dartmus]

I don't understand this.
Let's say I give you 100 entries and exits, each time at the top and bottom of each wave. Can you generate for me the reversed engineered code for that?
I tried Artificial Intelligence years ago. But never ended up with any profitable system.
If reversed engineering would really be interesting, companies like Virtue or Medallion, who have the knowledge and the resources, would reverse engineer all the available systems. But they don't...
Example: if you use (hypothetically) 4 timeframes, in 2 timeframes 6 indicators with specific rules, and in 2 timeframes 3 other indicators with other rules, which even can change depending on other indicators (so interpretation would change according to these other indicators).
I would like to see who can reverse engineer that.
To me it looks impossible, because the revere engineer has no clue about number of timeframes, length of each timeframe, which indicators, based on what and how to interpret...... You will have billions of theoretical possible combinations. So first find out how many combinations there are, and second how to eliminate and keep what is needed?

Investment funds don't send a list of yesterdays entries and exits to their competitors because analysis of their trades will reveal recurring anomalies that aren't present elsewhere. Whether reverse engineering relies on talent or artificial intelligence is a separate debate. There's multiple signals present at those locations that are not present in as strong of a confluence at similar locations where there aren't any trades. Code doesn't have to b an identical match. It can b quite different and focused on a different aspect of the signal.

I read where multiple programmers told u a discretionary portion of your method can't be automated. That's unlikely because it's dependent on mad skillz ...or ability to find a consistently recurring signal. It doesn't have to b the exact same signal and anomalies you're leaning on. They should have automated it and improved on what you're doing. In your case it might b difficult to improve your results but imo all rule based discretionary methods can be automated ...and improved.

 

[botpro]

By analysing the entry and exit signals one can calculate the stop% and target% used. This info is indeed useful.
But usually there is much more to a strategy, especially with complex basket/portfolio strategies,
so reverse engineering especially the entry points of such complex strategies is IMO impossible.

OTOH: if it is always just the same one instrument, and one collects much of the signals,
then one could try to come up with an equation or function that fits it best, ie. a best-fit approximation.

 

[i am nobody]

Investment funds don't send a list of yesterdays entries and exits to their competitors because analysis of their trades will reveal recurring anomalies that aren't present elsewhere. Whether reverse engineering relies on talent or artificial intelligence is a separate debate. There's multiple signals present at those locations that are not present in as strong of a confluence at similar locations where there aren't any trades. Code doesn't have to b an identical match. It can b quite different and focused on a different aspect of the signal.

I read where multiple programmers told u a discretionary portion of your method can't be automated. That's unlikely because it's dependent on mad skillz ...or ability to find a consistently recurring signal. It doesn't have to b the exact same signal and anomalies you're leaning on. They should have automated it and improved on what you're doing. In your case it might b difficult to improve your results but imo all rule based discretionary methods can be automated ...and improved.

I know that John Henry stopped giving realtime info to his clients, but it was because the competitors tried to force him out of a positions once they knew what he was doing. Had nothing to do with reverse engineering. Attack the opponent when he is at his weakest.

 

[userque]

By analysing the entry and exit signals one can calculate the stop% and target% used. This info is indeed useful.
But usually there is much more to a strategy, so reverse engineering especially the entry points is IMO impossible.
OTOH: if it is always just the same one instrument, and one collects much of the signals,
then one could try to come up with an equation or function that fits it best, ie. a best-fit approximation.

Then, you should do that using the local highs/lows of any instrument as your 'signals' and get rich.

 

[i am nobody]

By analysing the entry and exit signals one can calculate the stop% and target% used. This info is indeed useful.
But usually there is much more to a strategy, so reverse engineering especially the entry points is IMO impossible.
OTOH: if it is always just the same one instrument, and one collects much of the signals,
then one could try to come up with an equation or function that fits it best, ie. a best-fit approximation.

That is indeed the only thing possible.

But why do so many people never make money? They should just take imaginary entries and exits (the ones they dream about), and program a best-fit approach. Should be simple as cake if they believe in reverse engineering.

 

[i am nobody]

Then, you should do that using the local highs/lows of any instrument as your 'signals' and get rich.

I wrote the same as you.

 

[userque]

I know that John Henry stopped giving realtime info to his clients, but it was because the competitors tried to force him out of a positions once they knew what he was doing. Had nothing to do with reverse engineering. Attack the opponent when he is at his weakest.

Right. The OP is confusing protecting trading signals of traders with protecting what a fund will be buying over several days in a particular month.

 

[botpro]

That is indeed the only thing possible.

But why do so many people never make money? They should just take imaginary entries and exits (the ones they dream about), and program a best-fit approach. Should be simple as cake if they believe in reverse engineering.

Could be that they simply don't have the signals collected for RE. And how should they get that info and from where?