Kreslik.com - traders community since 2006

I have an automated strategy that I'm building.

Can I sell limit at the ask and buy limit at the bid successfully?
Any opinions? Or should I sell at the bid and buy at the ask?

It seems I can gain a pip by playing against the spread most of the time but what is the probability I won't get filled at all?

After watching the Forex Level II for many hours and trading it some manually, it appears that I can buy with a limit order at the bid and sell limit at the ask on most cases.

Years ago I traded the EMini successfully using limit orders and usually traded against the spread. Of course, that was counter trend trading.

I'm trying now to trade similar to the $Trillion strategy at crosses of the high and low of 10 minute bars. In many cases, this is going with the trend.

After all, surely trades occur at both the bid and ask prices due to market orders that come in constantly. I see the 2000 and 1000 mini lot order get whittled down at the bid and ask in just seconds must of the time.

However, sometimes there's a stall where the ask stops getting traded but the bid is getting traded more or vice versa.

Sincerely,
Wayne

What you are wanting to do is similar to trading inside the spread itself, also a method of market making. You can be successful at this if you pay very low/no commissions, you have a very quick line to the floor, and an itchy trigger finger. If you don't carry these 3 traits, I think you will find your account consistently eat up in 1 or 2 pip increments.

Also think of what could happen if you get your limit fill at the bid and the market doesn't let you out at the ask, soon you're down 4 or 5 pips which have to be made up. I wouldn't recommend this unless you were working on the floor.

Thanks jhtumblin.

But I wasn't clear. I wasn't refering to immediately buying and selling the bid/ask like you describe.

Instead, I was referring to intraday trading break out with limit orders. But should I set the limit order at the ask or bid when going long? Or vice versa on short?

Well, I have my own answer for anyone else who cares from both back and foward testing.

If your buying in auto trading you can set your limit order to the current ask price. Then if not filled within a specific number of seconds and you still want in (your conditions are still valid) chase the ask by increasing by a tick.

Personally, I found it most effective to set my limit price at 1 tick better than the ask. 60% of the time it fills at the ask but the rest of the time it either fills at a at my limit price OR even sometimes at a pip better.

If the market jumps past my limit, I will chase it by one pip. If still not filled , I cancel.

But when I tried testing by using the bid for long, it was never filled 50% of the time and the market ran away too quickly and it completely misses the trade.

Hope this helps.

Wayne

I imagine you mean that when buying you set a limit price to be the current bid? Buying at the current ask if like buying at market without slippage.

Basically, you're talking about a stop-limit order with a 1 tick buffer if the market doesn't tick back to the breakout price. Correct me if I'm wrong. Great idea, the trade-off you describe is the obvious downside - the market can (and will!) run away.

Gooni. Thanks for your comment. My idea really only applies to auto trading where the system can change the order in milliseconds. Faster than humans.

I'll probably just have to experiment with a live account to see.

Back when I traded the e-mini electronically. I always put in an order at the bid (if going long) first and if not filled simply bumped it up to the ask right away. More than 50% of the time, it goes filled at the bid price.

See? That makes another pip which adds up across many trades.

I'm thinking something advance like the auto system places the limit at the bid and sets a 100 milliseconds timer. And see if it gets filled.

NOTE: This would only be a test strategy on a micro lot.

IF it gets filled, it could try to exit the same way by selling at the ask with a 100 ms timer. If not filled, it could drop to the bid.

The test strategy could limit to when the market is "ranging" or "channeling" with zero momentum.

Next, if that doesn't work, the time could be gradually increased to find out how long it takes to get filled at the ask or bid over 50% of the time.

If it's only 100 milleseconds or less to get filled more than 50% of the time. Wow. that could add a lot of pips to a strategy.

Consider it "reverse slippage".

Sincerely,
Wayne

Yes, it's a very good idea and controlling these 'costs' of doing business certainly adds to your bottom line and separates the pros from the amateurs. Pulling it off successfully and automatically is a different story - it takes both excellent programming skills and good technical knowledge and hardware to get that kind of millisecond control over your system.

Thanks gooni, I agree with that in general. But I personally deal with far more challenging technical and performance issues in my day job. Writing the code for a trading system is the least difficult part from my perspective. It's finding and testing profitable rules that prove most challenging.

Interesting that you mention hardware. That's my weak point at the moment. Running this strategy from my home PC will be untenable over time even with a DSL connection since it has outages on occasion.

I clearly need a dedicated server. I've only priced rackspace.com so far and they want $350 a month. But they guarantee (and delivery) 100% network uptime and monitor the PC to make sure everything is running.

Any suggestions anyone?

I'm used to Linux/Unix hosting prices--much more affordable. But the MB Trading API and NinjaTrader only run under Windows. So I need a dedicated XP server.

Sincerely,
Wayne