Test a trading signal based on the closing price

[ajensen]

Daily open-high-low-close data is free, plentiful, and fast to test. But if your signal is generated using the closing price, how should you test it, without using intraday data? I'm not sure, but here are some thoughts.

(1) If your system is anti-trend, such as buy at the close when it is below the N-day moving average, you can calculate the price that triggers a signal and place a limit-on-close order. But if you enter a big order, couldn't your limit order affect the closing price?

(2) If you will calculate your signal 5 minutes before the close, you can test your signals on historical closing data with noise added, having standard deviation equal to the SD of returns from from close-5-minutes to close.

(3) You can trading on the next open or the next close. Performance may be degraded for a short-term system, but it's better to err on the conservative side.

Of course, it is best to buy and test with intraday data, but if you have 5 years of intraday data and 20 years of daily data, it's good to use all 20 years of data in some way.

 

[nooby_mcnoob]

Maybe controversial: it doesn't matter since the time between close and just before close is usually noise. Just add some slippage either way.

 

[lindq]

Daily open-high-low-close data is free, plentiful, and fast to test. But if your signal is generated using the closing price, how should you test it, without using intraday data? I'm not sure, but here are some thoughts.

(1) If your system is anti-trend, such as buy at the close when it is below the N-day moving average, you can calculate the price that triggers a signal and place a limit-on-close order. But if you enter a big order, couldn't your limit order affect the closing price?

(2) If you will calculate your signal 5 minutes before the close, you can test your signals on historical closing data with noise added, having standard deviation equal to the SD of returns from from close-5-minutes to close.

(3) You can trading on the next open or the next close. Performance may be degraded for a short-term system, but it's better to err on the conservative side.

Of course, it is best to buy and test with intraday data, but if you have 5 years of intraday data and 20 years of daily data, it's good to use all 20 years of data in some way.

Nothing wrong with using daily data. Many great systems are developed without access to intraday data for testing. Just be aware that for many instruments, the open/close is often more accurate that hi/lo.

If your system asks you to enter at the close, then your challenge is to locate a broker that will offer you a market-on-close option, or the option to set a time to enter the trade just prior to the close. In that case, you'll need to include the possibility of a little slippage at the entry.

Assuming that the open of the next daily bar is going to give you anywhere near the same results is a bad assumption.

Good luck.

 

[murray t turtle]

Maybe controversial: it doesn't matter since the time between close and just before close is usually noise. Just add some slippage either way.

%%
Practical points.And like the founder of Investors Business Daily says''DO NOT quibble over quarter + miss the move.''

It sure can pay to exit a trend late, especially a good trend, but that depends on the market also.Unless you are a market maker .......more has been made off a calendar than a clock.

A good trend, based off end of day data signal ;signal means do NOT miss the move.With out getting to exact,5 minutes may not be enough time for me to get in, even using end of day data, which i use.Besides ,since last 5 minutes can be super busy; dont miss the move. I may miss the clock, but dont miss the move.
Many mutual funds /funds have some legal requirements based on market close[3:00cst/4:00est]; but i dont see any of those on this thread yet.

 

[tiddlywinks]

IMO, the difference between signals off of Daily data versus fast intraday data, is drawdown. The slower the data, the larger the risk in terms of drawdown that must be allowed, WITHOUT negation of the signal. Negation is negation... reverse, or exit to the sideline.

A suggestion... Your analysis might benefit from using some form of volume analysis... market close and market open usually offer large(r) volumes, and if you can discern sustained price movement on rising, level, or declining volume, it may offer clues, or even tells. Just saying. If there are a million traders, there's 999,900 different ways to profit. To each their own.

 

[lindq]

IMO, the difference between signals off of Daily data versus fast intraday data, is drawdown. The slower the data, the larger the risk in terms of drawdown that must be allowed, WITHOUT negation of the signal. Negation is negation... reverse, or exit to the sideline.

No, if my testing on daily bars tells me that I should exit at X price within the bar, then I don't need intraday data in backtesting to set that stop at my broker when trading.

My best systems have been developed with daily data. I prefer to enter at an open, set a stop, and hold till the close. Either short of long.

Best move I ever made years ago was dumping my intraday testing.

 

[tiddlywinks]

No, if my testing on daily bars tells me that I should exit at X price within the bar, then I don't need intraday data in backtesting to set that stop at my broker when trading.

My best systems have been developed with daily data. I prefer to enter at an open, set a stop, and hold till the close. Either short of long.

Best move I ever made years ago was dumping my intraday testing.

This may have to do with your purpose for determining entry signals from daily data. For me, using slower bars is an attempt to capture longer-term (as compared to fast intraday) moves.

Something I learned long ago is to stay in my lane. For example, If trading intraday, a daily 200ma has no use to me, just as a one minute bar has no use to me if I am trading for longer-term movement. Sure, an argument could be made that faster is included within slower, thereby offering early detection, but that's not the way my methods work. To each their own.

Good trading to you.

 

[murray t turtle]

No, if my testing on daily bars tells me that I should exit at X price within the bar, then I don't need intraday data in backtesting to set that stop at my broker when trading.

My best systems have been developed with daily data. I prefer to enter at an open, set a stop, and hold till the close. Either short of long.

Best move I ever made years ago was dumping my intraday testing.

%% I see your good points.
And if one can find something the market really loves/good trend , or really hates/trend ; weekly charts can help a lot......... DrawDowns are a different thread; QQQ drawdowns are better than a loss.
I also see Tdd Win's points. Much of the money that has been lost, has been lost off intraday 5 minutes charts/noise. Market makers do real well off intra-day data,+ more power to them.

With discretion, i may miss the clock, but dont miss the move.Since i get signal off daily charts + some other stuff, but seldom do last minute MOC, have to enter off daily+intra-day, to be precise. But intra-day entry off, a daily chart signal/discretion is rightly named ,end of day signal .- ''dont quibble over quarter miss the move''

By the way, Don Bright DayTrading Co , did not like 5 minute charts; too slow.LOL+true.

 

[murray t turtle]

[Partial]QUOTE="tiddlywinks, post: 4799913, member: 69812"]IMO, the difference between signals off of Daily data versus fast intraday data, is drawdown. The slower the data, the larger the risk in terms of drawdown that must be allowed........

A suggestion... Your analysis might benefit from using some form of volume analysis... market close and market open usually offer large(r) volumes, and if you can discern sustained price movement on rising, level, or declining volume, it may offer clues, or even tells. Just saying. If there are a million traders, there's 999,900 different ways to profit. To each their own.[/QUOTE]
%%
I like to learn+ study trends; i just learned T Crabel [Crabel Capital Management]was born on 5-5-1955, today. I clipped a magazine article of his, not to long ago. NOT many will manage billions + average better than 16%, long term track record but he has; short term trading.
GM + DAL drawdowns went into bankruptcy court;+ plenty of stuff/ETFs........ has not gone into bankruptcy.

Most of the money made in US stock market, past 200+ years has been off draw downs, most likely. Prices/trends are subject to change. I liked the T Crabel short term article ; i remember he wanted everybody getting good execution prices......................................................................................................