Support Board
Support Board
Search Board
Control Panel
View My Posts / Threads
Direct MessagesDate/Time: Fri, 26 Apr 2024 13:44:37 +0000
Sierra Support: Too much difference between back-tested and live results
View Count: 1969
| [2019-01-25 14:47:47] |
| rahul - Posts: 160 |
|
Hello, I am using Real-Time Sierra Chart Exchange feed along with CQG WebAPI as a trading service. I have noticed the back-testing results is very different from live results. There seems to be a lot of slippage on order fills. The back-testing results are always more favorable than real-time results. My current setup: Sierra Chart Package: Level 3 Symbol Traded: ES, NQ (CME futures) When traded: Traded on Bar Close Order Type: Market Order Relevant code in Study: sc.UpdateAlways = 1 (so study is always called at chart update interval, even though there is no new market data) Back-Testing Settings: (Instance of Sierra Chart on computer 1) Service: SC Data - All Services Intraday Data Storage Time Unit: 1 Second Number of Stored Time and Sales Records: 4000 Allow Support for Sierra Chart Data Feeds: True (Checked) I am using "Auto Trade System Back Testing" using "Accurate Trading System Back Test Mode" Chart Update Interval: 200ms Real-Time Settings: (Separate instance of Sierra Chart on computer 2) Service: CQG WebAPI Intraday Data Storage Time Unit: 1 Second Number of Stored Time and Sales Records: 4000 Allow Support for Sierra Chart Data Feeds: TRUE (CHECKED) Chart Update Interval: 200ms I can see [M] next to the CQG WebAPI symbol, so Real-Time Sierra Chart Exchange feed is working My questions are around how I can get a closer values of back-testing and real-time trading? 1. Should I change the Intraday Data Storage Time Unit = 1 Tick for the Back-Testing? 2. Should I be back-testing using the Real-Time settings instead but with Simulated Mode on? Should I change this value to 1 Tick or keep it at 1 second? 3. Should I increase the chart update interval for real-time trading? If yes, to what value? 4. Should I change my trading service to CQG FIX or something else? 5. The documentation says that "You will have the greatest consistency between back testing and real-time auto trade system evaluation when you are performing a Replay Based Back Test on tick by tick data". Since the replay method is very slow, what's the fastest speed I can replay at to still maintain accuracy? Date Time Of Last Edit: 2019-01-25 19:48:10
|
| [2019-01-28 17:30:23] |
| |
|
There will be differences. Refer to: Auto Trade System Back Testing: Differences Between Back Testing and Real-Time Auto Trade System Evaluation 1. Should I change the Intraday Data Storage Time Unit = 1 Tick for the Back-Testing? 2. Should I be back-testing using the Real-Time settings instead but with Simulated Mode on? Should I change this value to 1 Tick or keep it at 1 second? Use 1 Tick.3. Should I increase the chart update interval for real-time trading? If yes, to what value? No we do not think that you need to do that.4. Should I change my trading service to CQG FIX or something else? This is irrelevant.5. The documentation says that "You will have the greatest consistency between back testing and real-time auto trade system evaluation when you are performing a Replay Based Back Test on tick by tick data". Since the replay method is very slow, what's the fastest speed I can replay at to still maintain accuracy? The speed is not relevant so long as you are using Accurate Trading System Back Test Mode:https://www.sierrachart.com/index.php?page=doc/ReplayChart.html#ReplayMode Sierra Chart Support - Engineering Level Your definitive source for support. Other responses are from users. Try to keep your questions brief and to the point. Be aware of support policy: https://www.sierrachart.com/index.php?l=PostingInformation.php#GeneralInformation For the most reliable, advanced, and zero cost futures order routing, *change* to the Teton service: Sierra Chart Teton Futures Order Routing |
| [2020-09-10 20:15:01] |
| User19107 - Posts: 7 |
|
rahul: Did you make any progress with this?
|
| [2020-09-10 22:36:00] |
| bradh - Posts: 854 |
|
One important thing to consider with live vs. replay is the latency (ping time) between you and the exchange. In live trading (or live sim) you get a delay seeing the data, and another delay sending your order. The price has had 2 x the delay time to move from where you think it is. In Replay, there is no delay seeing the data, as it is being played back from a file. When you see it price hasn't moved yet. When you submit your order, the simulated exchange in Sierra responds as fast as it can, consistently. (Unless there is some deliberate delay programmed into the simulator, but I don't think there is one.) Since replay doesn't simulate your actual round trip delay to the exchange, you are subject to more slippage. Using limit orders instead of market orders can help the predictability of your back tests, as the are on the server at the exchange. |
To post a message in this thread, you need to log in with your Sierra Chart account: