No not using the IB excel DDE API. After the studies are run for the next day's calls, trades are manually entered in TWS using the time variable to trigger the execution during pre-market. Again, will need to research if I can tie the decision studies into the Excel DDE API to create entry automation.
2018 - 8 Future Basket Trading Journal
- Thread starter MidwesternTrader
- Start date
Trading through Drawdowns. Grinding Until Recovery.
This topic is what I am focused on this week. Can I properly execute my strategy calls after a draw down week? I call it “The Grind”. When you have a strategy or any other approach with a positive expectancy over time, are you able to grind trough drawdowns until positive results return? Especially if the drawdown has caused a lack of confidence or an emotional downer due losses. “The Grind” is a process every trader will need to master. And it applies to every type of trader – fully automated program trading, system/strategy calls, 100% discretionary trading and gunslingers alike.
When people back test trading ideas, everyone is familiar with drawdown and maximum drawdown. The maximum drawdown being the largest peak to valley drop in the P & L curve during the period back tested. But how does the strategy performs during the recovery periods? Recovery periods are the number of trades (or days) it takes to fully recover the draw down. As this strategy trades 8 instruments every day, I measure its recovery period in days.
When I am exploring strategies, I like to see quick recovery periods. I favor trying strategies with the ability to “snap back” to profitability quickly. For the 8 Futures Basket Strategy using a $50,000 starting account, I consider any draw down in excess of -$4,000 a major draw. I want to know how many major drawdowns occur and, just as importantly, how many trading days it takes to fully recover each drawdown.
The table below lists each major draw and the number of days it took to fully recover the draw before the strategy’s P & L mountain climbed to a new high. This data set is from 2016 - 2017. It includes commissions and trade slippage.
When developing trading strategies, having a low drawdown-recovery ratio is highly preferred. Knowing this average ratio can help one understand how long it will take to “Grind” back to profitability. This is another factor that should be considered when developing trading strategies and systems that some overlook.
This topic is what I am focused on this week. Can I properly execute my strategy calls after a draw down week? I call it “The Grind”. When you have a strategy or any other approach with a positive expectancy over time, are you able to grind trough drawdowns until positive results return? Especially if the drawdown has caused a lack of confidence or an emotional downer due losses. “The Grind” is a process every trader will need to master. And it applies to every type of trader – fully automated program trading, system/strategy calls, 100% discretionary trading and gunslingers alike.
When people back test trading ideas, everyone is familiar with drawdown and maximum drawdown. The maximum drawdown being the largest peak to valley drop in the P & L curve during the period back tested. But how does the strategy performs during the recovery periods? Recovery periods are the number of trades (or days) it takes to fully recover the draw down. As this strategy trades 8 instruments every day, I measure its recovery period in days.
When I am exploring strategies, I like to see quick recovery periods. I favor trying strategies with the ability to “snap back” to profitability quickly. For the 8 Futures Basket Strategy using a $50,000 starting account, I consider any draw down in excess of -$4,000 a major draw. I want to know how many major drawdowns occur and, just as importantly, how many trading days it takes to fully recover each drawdown.
The table below lists each major draw and the number of days it took to fully recover the draw before the strategy’s P & L mountain climbed to a new high. This data set is from 2016 - 2017. It includes commissions and trade slippage.
- -$5,250 : Trading Days to Recover : 6
- -$6,230 : Trading Days to Recover : 6
- -$7,680 : Trading Days to Recover : 8
- -$4,460 : Trading Days to Recover : 2
- -$7,680 : Trading Days to Recover : 8
- -$4,290 : Trading Days to Recover : 1
- -$7,750 : Trading Days to Recover : 13
- -$4,030 : Trading Days to Recover : 4
- -$4,330 : Trading Days to Recover : 3
- -$5,780 : Trading Days to Recover : 5
- -$4,000 : Trading Days to Recover : 12
- -$6,040 : Trading Days to Recover : 7
- -$5,700 : Trading Days to Recover : 4
- -$4,690 : Trading Days to Recover : 1
- -$4,390 : Trading Days to Recover : 2
- -$7,080 : Trading Days to Recover : 5
- -$9,200 : Trading Days to Recover : 4
When developing trading strategies, having a low drawdown-recovery ratio is highly preferred. Knowing this average ratio can help one understand how long it will take to “Grind” back to profitability. This is another factor that should be considered when developing trading strategies and systems that some overlook.
Week Jan 22nd – Jan 26th
Back to green, but still in “The Grind”. The week was a good net positive, but still grinding away until the P & L curve is able to recover the losses from the previous week. Did a good job of executing all signals as presented. Back testing from 2016 – 2017 for this strategy showed only two instances of back-to-back losing weeks, so this recovery week met historical expectations.
3 wining days, 2 losing days. 21 winning trades, 19 losing trades with 13 full stop losses.
Note the percentage of winning vs. losing trades was 53% - 47%, nearly 50 – 50. The practice of using strict stops for losers and letting winning trades run was the primary driver for a positive week. Four out-sized gainers account for the week’s net profits. The terms “always use stop losses” and “proper money management” are often just highly repeated phrases in the trading world. But when applied and executed properly in live trading, they will improve performance over the long term.
NAV: +$3,690.35
Adherence to outlined trading strategy: 100%.
Finally, I will be adjusting the KC (Coffee) trade to a wider stop loss level by moving from a -1.5 cent stop to the -3.0 stop. The historical, total P & L results for both stop levels are roughly the same, but live trading the tighter -1.5 stop is showing cracks due low liquidity of this instrument. I’ll post a short analysis piece of the issues that led to this decision.
Back to green, but still in “The Grind”. The week was a good net positive, but still grinding away until the P & L curve is able to recover the losses from the previous week. Did a good job of executing all signals as presented. Back testing from 2016 – 2017 for this strategy showed only two instances of back-to-back losing weeks, so this recovery week met historical expectations.
3 wining days, 2 losing days. 21 winning trades, 19 losing trades with 13 full stop losses.
Note the percentage of winning vs. losing trades was 53% - 47%, nearly 50 – 50. The practice of using strict stops for losers and letting winning trades run was the primary driver for a positive week. Four out-sized gainers account for the week’s net profits. The terms “always use stop losses” and “proper money management” are often just highly repeated phrases in the trading world. But when applied and executed properly in live trading, they will improve performance over the long term.
NAV: +$3,690.35
Adherence to outlined trading strategy: 100%.
Finally, I will be adjusting the KC (Coffee) trade to a wider stop loss level by moving from a -1.5 cent stop to the -3.0 stop. The historical, total P & L results for both stop levels are roughly the same, but live trading the tighter -1.5 stop is showing cracks due low liquidity of this instrument. I’ll post a short analysis piece of the issues that led to this decision.
Making Adjustments After Going Live – KC (Coffee) Trade
This is an example of adjusting a trading plan after experiencing live results.
Simulated and back tested trading results assume perfect entries and exits. It is normal practice to assign a slippage factor to a back test, but real slippage is unknown until live trading begins. My back test results over 2016 – 1017 on KC (Coffee) showed slightly better P & L performance using a 1.5 cent stop versus a 3.0 cent stop. Those equate to a -$562.50 and -$1,125 respectively as KC futures are multiplied by $375 per cent. I originally chose to use the -1.5 cent stop due to the lower loss level on full stop loss trades..
KC is a thinly traded instrument, so it is more prone to higher slippage impact than say NQ. During the first six weeks of live trading, the low liquidity in KC has revealed a major slippage impact on the -1.5 cent stop loss exits. Once or twice a week, KC will experience a stop loss “blow out” that sends the instrument sharply higher or lower in under 10 seconds. The 5 minute and 1 minute charts below show this blow out price action from 1-25-18. This 1.4 cent drop in KC would be the equivalent of the DOW futures dropping 500 points in 10 seconds.
5 MIN KC Chart 1-26-18
1 MIN KC Chart 1-26-18
The price movement in this blow out was 1.4 cents in under 10 seconds. Stop loss orders trigger during this price drop were filled significantly lower than the stop price trigger.
One possible solution is to use a Stop with Protection order as that order type is native to the ICE exchange. This order type for a long position sets a Stop amount to trigger a sell at market order and also includes a price limit that signifies the lowest price that will be accepted for a fill. As the order is native to ICE, it may improve the order’s positioning allowing the exit closer to the desired stop amount. But if the blowout begins ahead of the stop price trigger, other stop orders could still be queued in front of your Stop Limit order causing price to fall entirely past your limit price. This would cause a "no fill"l until price moved back up (if at all) to your limit price designation. This presents the risk of not receiving a fill when using the Stop with Protection order (Stop Limit).
I researched the price blowouts for KC in more detail over the past two years. They occur on average once or twice a week with about 75% of them within +/- 2 cents of the 4AM CST opening trade. My thought of using the -3 cent is intended to have the stop outside the price zones where these blowouts typically occur. So going forward I adjust the trading plan to use the -3 cent stop using normal Stop Loss orders for KC. There will still be instances where blowouts will impact the -3 cent stop, but hopefully far fewer times than the -1.5 cent stop loss level. I’ll continue using the regular Stop Loss order instead of the Stop with Protection as I feel the risk of not getting filled to close the position in a major market move is greater than taking on additional slippage during the “blow outs”.
This is an example of adjusting a trading plan after experiencing live results.
Simulated and back tested trading results assume perfect entries and exits. It is normal practice to assign a slippage factor to a back test, but real slippage is unknown until live trading begins. My back test results over 2016 – 1017 on KC (Coffee) showed slightly better P & L performance using a 1.5 cent stop versus a 3.0 cent stop. Those equate to a -$562.50 and -$1,125 respectively as KC futures are multiplied by $375 per cent. I originally chose to use the -1.5 cent stop due to the lower loss level on full stop loss trades..
KC is a thinly traded instrument, so it is more prone to higher slippage impact than say NQ. During the first six weeks of live trading, the low liquidity in KC has revealed a major slippage impact on the -1.5 cent stop loss exits. Once or twice a week, KC will experience a stop loss “blow out” that sends the instrument sharply higher or lower in under 10 seconds. The 5 minute and 1 minute charts below show this blow out price action from 1-25-18. This 1.4 cent drop in KC would be the equivalent of the DOW futures dropping 500 points in 10 seconds.
5 MIN KC Chart 1-26-18
1 MIN KC Chart 1-26-18
The price movement in this blow out was 1.4 cents in under 10 seconds. Stop loss orders trigger during this price drop were filled significantly lower than the stop price trigger.
One possible solution is to use a Stop with Protection order as that order type is native to the ICE exchange. This order type for a long position sets a Stop amount to trigger a sell at market order and also includes a price limit that signifies the lowest price that will be accepted for a fill. As the order is native to ICE, it may improve the order’s positioning allowing the exit closer to the desired stop amount. But if the blowout begins ahead of the stop price trigger, other stop orders could still be queued in front of your Stop Limit order causing price to fall entirely past your limit price. This would cause a "no fill"l until price moved back up (if at all) to your limit price designation. This presents the risk of not receiving a fill when using the Stop with Protection order (Stop Limit).
I researched the price blowouts for KC in more detail over the past two years. They occur on average once or twice a week with about 75% of them within +/- 2 cents of the 4AM CST opening trade. My thought of using the -3 cent is intended to have the stop outside the price zones where these blowouts typically occur. So going forward I adjust the trading plan to use the -3 cent stop using normal Stop Loss orders for KC. There will still be instances where blowouts will impact the -3 cent stop, but hopefully far fewer times than the -1.5 cent stop loss level. I’ll continue using the regular Stop Loss order instead of the Stop with Protection as I feel the risk of not getting filled to close the position in a major market move is greater than taking on additional slippage during the “blow outs”.
What is your slippage estimate vs profit expectation for trading one lot of KC per year?
I mean, lets say the back test says you would of made 20K a year trading one KC lot.
How much would you be prepared to lose in slippage before you would remove KC from the basket on the basis it is not suitable for day trading due to these frequent nasty spikes?
For each instrument I assigned a base slip equivalent to one tick considering 50% of the time you would get the same fill as the bar close print price and 50% of the time you would get the filled on the other side of the bid/ask than the bar close print. This slip assignment seems to be accurate for highly liquid instruments like CL and NQ. But it is logical that less liquid instruments are more prone to additional slippage. For KC one tick is $18.75 (5/100 * $375). The average yearly return on KC for 2016 and 2017 (including one tick slip per trade) was $37,460 per year.
During the six weeks of live trades (Dec 18th - Jan 26th), the strategy's baseline expected KC return with the one tick slippage is -$1,642. The actual traded NAV for that time has been -$2,579. Therefore live trading has resulted in additional slippage of -$937 over all trades or an average of -$31 more slip per trade than the back tested expectancy. This amount equates to roughly the additional slip caused by the nasty spikes when my stop order has been caught in them.
If this additional average slippage would occur for an entire year (260 trades), the average expected KC return would be reduced from $37,460 to $29,400. Still worthwhile to be traded. But going live on the -3.0 stop for 6 weeks will provide some comparison as to whether the "price spike" slippage can be mitigated.
Also, FWIW KC has the highest average expected return of any of the 8 futures contracts during the 2 year back test. My guess is this year would be a "reversion to the mean" year for KC and show a lower total return than the two previous years. So finding a alternative stop loss level able to mitigate "spike" price slippage is definitely preferable. I'll just need to live trade the -3.0 stop loss level and see if additional slippage continues or not.
I would say that slippage number matches to my first 8 weeks of KC live trading.
Question to the crowd - When to roll trading forward financially settled contracts?
For deliverable contracts, IB shuts down buying and selling a couple days before the first position date, so that is when I move to the next contract month to trade CL, GC, NG, HO, and LE.
For NQ I'll roll forward about a week before the final trade date.
But does anyone have a method or reasoning they care to share for when to roll forward for KC and HE? My thought is to move to the next month as its volume begins to outpace the current month, but wondered if I should look at anything else.
For deliverable contracts, IB shuts down buying and selling a couple days before the first position date, so that is when I move to the next contract month to trade CL, GC, NG, HO, and LE.
For NQ I'll roll forward about a week before the final trade date.
But does anyone have a method or reasoning they care to share for when to roll forward for KC and HE? My thought is to move to the next month as its volume begins to outpace the current month, but wondered if I should look at anything else.
For US SIFs like NQ and ES, the convention seems to be to switch on the Friday before expiry.
For the European SIFs i trade the convention seems to be to wait until the Expiry day it self.
I don't trade the other markets you trade, but there is probably a standard fixed number of days before expiry when most traders switch to the next contract for each market. If you look at the volume data you can probably figure it out.
For the European SIFs i trade the convention seems to be to wait until the Expiry day it self.
I don't trade the other markets you trade, but there is probably a standard fixed number of days before expiry when most traders switch to the next contract for each market. If you look at the volume data you can probably figure it out.