Spacious Spread

Strategies might be attractive on the surface but fail moving forward. Why? Here we discuss the qualities of effective trading strategies, and the many traps one should avoid.

Spacious Spread

Postby Dacamic » Fri Apr 07, 2006 3:27 pm

My systems always include trading rules designed to keep me away from illiquid securities. I nonetheless occasionally get signals for low-volume stocks ... which have monstrous spreads. For example, I tracked a stock yesterday that had a 4% spread between its bid and ask. When such ugliness occurs, we have a few options to consider:

-- Place a market order;
-- Place a limit order; and,
-- Wait and watch.

My systems include sufficient activity to avoid the need to dwell intently on any one trade. Regardless, I couldn't get comfortable yesterday leaving 4% on the table. So, rather than using my typical "market order" approach, I used a limit order to execute my trade. As it worked out, the trade executed at a price equal to the open price and 3% lower than the initial ask price. My joy knew no bounds.

We stand on a slippery slope when we deviate from assumptions used in our back tests. One exception becomes two ... then three ... then thirty. If we're not careful, we eventually rationalize ourselves away from profitable systems we carefully nurtured through the development process. In my circumstances yesterday, the textbook-correct method was to place a market order prior to the open, then let nature take its course. In this case, nature would have almost certainly executed my order at the ask price. That was an outcome sufficiently unpleasant to prompt my interference with the natural order ... and risk a step down the slope.

We might quickly spin into a circular argument trying to figure whether my actions violated basic assumptions in my back tests, regardless of their favorable outcome. My advice in this regard is to not let extraordinary steps become the norm. If I continue to bump into illiquid issues, I will run back tests with more restrictive volume filters. In the meantime, more head scratching will probably be needed when it comes time to exit the quiet little stock my system found yesterday.
Dacamic
 
Posts: 457
Joined: Wed Nov 30, 2005 12:40 pm

Postby Overload » Mon Apr 10, 2006 12:05 pm

Do you recall where you placed your limit, relative to the bid/ask or prior day's close?

Whether to use limit or market orders is an interesting question. I have largely used market orders, but the stocks I've traded (with my existing systems) have been very high volume. So there hasn't been much chance of catastrophe. On the other hand, it's not uncommon even with high volume stocks to see someone stick a lowball or highball offer out there. For example, the stock trades around $50, but someone's placed an order to sell at $100. I've never been hit on the losing side of one of those yet, but it's something to be aware of. If you do a market order, you're going to get the best price available at the time. But if nothing but highball/lowball offers are remaining out there, those will be what you get.

Since you're absolutely right that one must simulate the StrataSearch back tests as closely as possible to get the simulated returns, one must get as close the opening price as possible. And market orders can get you closer to that than limit orders. But... as mentioned above, there's a hidden trap doing that.

Perhaps the use of limit orders with an assumed slippage would be best. After trading limit orders for a period, you can then get a better idea of what slippage occurs within the sectors you're trading, and adjust back testing (and expectations) accordingly.

Pete
Overload
 
Posts: 2248
Joined: Wed Nov 30, 2005 12:14 pm

Postby Dacamic » Tue Apr 11, 2006 11:30 am

Overload wrote:Do you recall where you placed your limit, relative to the bid/ask or prior day's close?

... Perhaps the use of limit orders with an assumed slippage would be best. After trading limit orders for a period, you can then get a better idea of what slippage occurs within the sectors you're trading, and adjust back testing (and expectations) accordingly.

Pete


I placed my limit order at the previous day's close, which was about 1% above bid ... yet still 3% below ask. If my trade had not executed at that price, I was willing to chase the trade until prices rose 1% above the open.

I do not have much experience trading low-volume stocks, so my back tests exclude this portion of the trading universe. It seems, though, assumptions may become increasingly fragile as trading volume declines. There are more trapdoors in this arena than I care to cope with. That's just me, though.

My back tests are based upon market orders, and I use market orders almost exclusively for placing trades. This set-up has worked sufficiently well to allow my trades to execute with the slippage and spread assumptions used in back tests. The choice to use a limit order in this particular case was based upon circumstances anomalous to my system. If such anomalies become the norm (normal anomalies ... oxymoron?), it's more likely I'll adjust my volume filters than switch to limit orders. There ain't nuttin' wrong with limit orders; they just don't fit well into my current system.
Dacamic
 
Posts: 457
Joined: Wed Nov 30, 2005 12:40 pm

Slippage?

Postby kdstrategyseek » Sun Jul 16, 2006 6:49 pm

1) What slippage Percent amount do you set up in backtest? Spread Size Points defaults to 0.1.

2) There's a feature in Stratasearch that you can set up as limit order for backtest instead of using market order. Have you tried it?
kdstrategyseek
 
Posts: 20
Joined: Sun Jun 04, 2006 10:10 pm

Re: Slippage?

Postby Dacamic » Mon Jul 17, 2006 2:05 pm

kdstrategyseek wrote:1) What slippage Percent amount do you set up in backtest? Spread Size Points defaults to 0.1.

2) There's a feature in Stratasearch that you can set up as limit order for backtest instead of using market order. Have you tried it?


I use slippage and spread settings of 0.5% and $.07, respectively, which reflect my historical results.

I rarely use StrataSearch's limit order rules in my back tests. Limit orders have not added value to my live trading, so I have a slight negative bias against using them in back tests. Our users expressed an interest in exploring limit order-based systems, which prompted us to include the related functionality in Version 3.0. My personal preferences (and biases) have steered me away from limit orders; obviously, other people have found them worth considering for their systems. To each their own.
Steve
Dacamic
 
Posts: 457
Joined: Wed Nov 30, 2005 12:40 pm


Return to Curve-Fitting and Other Pitfalls