Are my eyes deceiving me?!!!

vikana · Jun 11, 2003

A simple idea is to test entry and exit on the worst possible price on the next bar. For a long entry, enter at the high of the next bar, and exit on the low of the exit bar. If your system can stay profitable with that, you should be ok.

rtstrading · Jun 11, 2003

What is the points/trade for the system w/o any commish
/slippage?
It needs to be at least 1.5-2.0 pts/trade to account for the 0.50-1.00 pt/trade slippage that you will encounter in realtime.

Shoot for systems that make at least 2 pts/trade for the NQ.

Good luck

Bob

catmango · Jun 11, 2003

Quote from vikana:

A simple idea is to test entry and exit on the worst possible price on the next bar. For a long entry, enter at the high of the next bar, and exit on the low of the exit bar. If your system can stay profitable with that, you should be ok.

That sounds pretty extreme, particularly for a scalp. Wouldn't it be better to say that for a long entry, if the open of the next bar is lower, then use the close of the previous bar, but if the open is higher, than assume the open? In other words, use the higher of the two for the buys and the lower of th two for the sells?

In terms of execution, this would require placing a market order right at the close, which is the approach that peterfigliozzi uses.

axeman · Jun 11, 2003

Simply dont assume you will get a fill unless it prints at least
1 tick through the price you want.

Unfortunately, this kills most scalping systems in backtesting.

You have to trade it live to know if you get hit enough for
it to work.

peace

axeman

spreadem · Jun 11, 2003

In terms of execution, this would require placing a market order right at the close, which is the approach that peterfigliozzi uses.

If using market orders, add 2 ticks to all buys and subtract 2 ticks from all sells.

bidmasterx · Jun 11, 2003

Quote from axeman:

Simply dont assume you will get a fill unless it prints at least
1 tick through the price you want.

Unfortunately, this kills most scalping systems in backtesting.

You have to trade it live to know if you get hit enough for
it to work.

peace

axeman

I think axeman is absolutely correct. The way I determine if I get a fill when testing a system is to see if there is a print that is at least 1 tick through my order price. I'm wondering why spreadem uses 2 ticks, I don't use market orders on GLOBEX so I'm no expert but isn't that being too conservative and not very accurate?

nitro · Jun 11, 2003

Quote from laziz314159:

I think smartquant can do this. I've heard rumours that the license is in the 2-3k range. Check the smartquant group on yahoo for more.

http://www.smartquant.com

K can do it too.. but good luck being able to afford it.. I've heard 200k for developer license (but that may have been a WAG).

http://www.kx.com

Don't forget how much the data will cost.. TAQ is not cheap.

--laz

Thanks for the links!!!! Never heard of either of these. kewl.

When I meant in existence, I meant "you can buy it" existence. The one we developed at the Hedgie I used to work at did this

nitro

trend456 · Jun 11, 2003

Quote from bidmasterx:

I think axeman is absolutely correct. The way I determine if I get a fill when testing a system is to see if there is a print that is at least 1 tick through my order price. I'm wondering why spreadem uses 2 ticks, I don't use market orders on GLOBEX so I'm no expert but isn't that being too conservative and not very accurate?

I've been using 2 ticks for ES on both sides. I'm sort of on the conservative side when it comes to trading systems. I experienced times when the market hits your exit but you don't get the fill. Your system records the last price. Shows a signal to exit and never hits the price again. When things like that occur I just quickly get out using MO's. I usually don't lose 2 ticks but I always want to consider the worst case scenario. I give it a lot of room.

It's all pretty much my opinion but I think it's more reliable if you consider more slippage than you expect.

Good Trade!!

trend

vikana · Jun 11, 2003

Quote from rolegario:

That sounds pretty extreme, particularly for a scalp. Wouldn't it be better to say that for a long entry, if the open of the next bar is lower, then use the close of the previous bar, but if the open is higher, than assume the open? In other words, use the higher of the two for the buys and the lower of th two for the sells?

In terms of execution, this would require placing a market order right at the close, which is the approach that peterfigliozzi uses.

Actually, thinking a bit more, you're probably right. The rule of using the worst price on the bars of entry/exit is what I use for market orders on the NYSE. Liquidy in ES/NQ would make 1 to 2 ticks enough to ensure a fill.

spreadem · Jun 12, 2003

Liquidy in ES/NQ would make 1 to 2 ticks enough to ensure a fill.

The main point of backtesting is not to find out the maximum amount of money that your system can make, it is to find out the minimum amount of money your system can make. In fact if you run the test under ideal circumstances and perfect fills, you'll be disappointed when the actual results come in.

I'd rather run my system tests under the assumption of getting poor fills and compare the actual results.

... Just my 2 cents ...