I received a detailed reply from buttontrader and I will post that below with their permission for the interest of the thread.
Ditch, I would not spend $50 for the answer of a question I can ask for free, whether I had $100 or $100,000. A tight arse? Absolutely.
Slow? Not usually. I also never said I was raking it in big on my swing trades. Thanks for your feedback regarding the simulator though.
Here is the response from buttontrader. I hope it is not breaching any rules of the forum. If it is I apologise in advance.
Hi Mark,
Thank you for the direct email. Someone of the Beta-testers noticed me of the ET-thread, but although I like to respond. I may not do that because the moderators will not allow someone from a commercial-product to use their forum (without paying them for it {which we will not do because all our resources go into development and not in marketing}).
Now to your question: who realistic is the Simulator?
This is actually a very difficult question so Iâm afraid I will need to give you a somewhat complex answer as well.
First it depends where you enter your Limit Buy Order:
1. Above or on the current Ask-price.
2. Below the Ask but on the Bid.
3. Below the Bid but in the MarketDepth region (5 ticks for ES)
4. Far below anything of above.
Only point-1 will immediately give an execution and your trade will be 100% realistic.
For points 2 till 4 there is a second condition: it depends on how long the waiting-row in the Order-book is, and where your position in the row is on a point in time.
In other words: your Order enters one of the waiting-rows (every tick-price has one) in the back and during trading-time your position is moving in this FIFO-row. Your actual question is: "how realistic can a simulation program like ButtonTrader determine on any moment of time where your Order is in the row".
Well no program can determine that with any realistic accuracy because needed information about the OrderBook and the modifications on that OrderBook are not present, not up-to-date or even available (point-4 of above).
For points-2 and â3, the IB data-stream only delivers the current length of the row, but these rows are changing every micro-second with new enters on the back, executions in front, and what is very common: pull backs in the middle of the rows. (The big guys regularly manipulate the OrderBook by entering very high volume lots and pulling them back before any execution, and then entering the same quantity back in the book: by this they can create a visible but fictive resistance level in the Book).
Bottom-line it is not possible to tell where your Order would be in the row, so also not when it executes in the simulator. The only way is to approach this by "probability calculations". And that is what ButtonTrader does. It tries to calculate the waiting-time that your order is in the row before your turn is on. I can not publicize a detailed explanation of this probably calculation but several dynamically changing parameters are involved like the Ask/Bid/Last and MarketDepth sizes and the volume and volatility in these sizes (and some more).
The result is that the simulator will execute your Order sometimes too quick and sometimes to slow compared with the reality, but that on average it will come close.
Back to your original question: is it realistic for ES with 1 to 3 Tick scalps.
Well to be honest: I doubt it. Trading 1 or 2 ticks on an OrderBook, which has nowadays more then 1000 lots per level will probably result in too many turnarounds that did get executions too soon. So to be at the safe side I would divide your simulated turnarounds by 2.
Why? You will ask. Well I made the "probability calculation" more then 2 years ago. At that time the NQ did 100 Lots and ES 400 Lots maximums (these days it is NQ 600 and ES 1500). And I tested the algorithm at that time in real trading and at the same time in simulation trading (you can do that in ButtonTrader: trading 2 strategies at the same time in Real and Simu in the same ButtonGrid to see the difference in executions)). The algorithm is tuned to that but now you have triggered me to look at it again. Probably I need to fine-tune the waiting parameters a bit to approximate the current market-situation.
Mark:
Long answer and I hope you appreciate it. Feel free to copy this answer to the ET forum if you think it is of interest by other users (As I said, I can not do that).
By the way: there is indeed a Yahoo-group (one of the Beta-testers was that enthusiastic to create it), but nearly all ButtonTrader users know that they get quick and honest response when they email us directly. So nearly all communication goes directly to us instead of through forums.
Hope this answers your question, otherwise please ask.
Robert
Ditch, I would not spend $50 for the answer of a question I can ask for free, whether I had $100 or $100,000. A tight arse? Absolutely.
Slow? Not usually. I also never said I was raking it in big on my swing trades. Thanks for your feedback regarding the simulator though.Here is the response from buttontrader. I hope it is not breaching any rules of the forum. If it is I apologise in advance.
Hi Mark,
Thank you for the direct email. Someone of the Beta-testers noticed me of the ET-thread, but although I like to respond. I may not do that because the moderators will not allow someone from a commercial-product to use their forum (without paying them for it {which we will not do because all our resources go into development and not in marketing}).
Now to your question: who realistic is the Simulator?
This is actually a very difficult question so Iâm afraid I will need to give you a somewhat complex answer as well.
First it depends where you enter your Limit Buy Order:
1. Above or on the current Ask-price.
2. Below the Ask but on the Bid.
3. Below the Bid but in the MarketDepth region (5 ticks for ES)
4. Far below anything of above.
Only point-1 will immediately give an execution and your trade will be 100% realistic.
For points 2 till 4 there is a second condition: it depends on how long the waiting-row in the Order-book is, and where your position in the row is on a point in time.
In other words: your Order enters one of the waiting-rows (every tick-price has one) in the back and during trading-time your position is moving in this FIFO-row. Your actual question is: "how realistic can a simulation program like ButtonTrader determine on any moment of time where your Order is in the row".
Well no program can determine that with any realistic accuracy because needed information about the OrderBook and the modifications on that OrderBook are not present, not up-to-date or even available (point-4 of above).
For points-2 and â3, the IB data-stream only delivers the current length of the row, but these rows are changing every micro-second with new enters on the back, executions in front, and what is very common: pull backs in the middle of the rows. (The big guys regularly manipulate the OrderBook by entering very high volume lots and pulling them back before any execution, and then entering the same quantity back in the book: by this they can create a visible but fictive resistance level in the Book).
Bottom-line it is not possible to tell where your Order would be in the row, so also not when it executes in the simulator. The only way is to approach this by "probability calculations". And that is what ButtonTrader does. It tries to calculate the waiting-time that your order is in the row before your turn is on. I can not publicize a detailed explanation of this probably calculation but several dynamically changing parameters are involved like the Ask/Bid/Last and MarketDepth sizes and the volume and volatility in these sizes (and some more).
The result is that the simulator will execute your Order sometimes too quick and sometimes to slow compared with the reality, but that on average it will come close.
Back to your original question: is it realistic for ES with 1 to 3 Tick scalps.
Well to be honest: I doubt it. Trading 1 or 2 ticks on an OrderBook, which has nowadays more then 1000 lots per level will probably result in too many turnarounds that did get executions too soon. So to be at the safe side I would divide your simulated turnarounds by 2.
Why? You will ask. Well I made the "probability calculation" more then 2 years ago. At that time the NQ did 100 Lots and ES 400 Lots maximums (these days it is NQ 600 and ES 1500). And I tested the algorithm at that time in real trading and at the same time in simulation trading (you can do that in ButtonTrader: trading 2 strategies at the same time in Real and Simu in the same ButtonGrid to see the difference in executions)). The algorithm is tuned to that but now you have triggered me to look at it again. Probably I need to fine-tune the waiting parameters a bit to approximate the current market-situation.
Mark:
Long answer and I hope you appreciate it. Feel free to copy this answer to the ET forum if you think it is of interest by other users (As I said, I can not do that).
By the way: there is indeed a Yahoo-group (one of the Beta-testers was that enthusiastic to create it), but nearly all ButtonTrader users know that they get quick and honest response when they email us directly. So nearly all communication goes directly to us instead of through forums.
Hope this answers your question, otherwise please ask.
Robert
But I suspected that from the start and why this thread was started. 