keohir808 has contributed to 6 posts out of 21083 total posts
(0.03%) in 1,470 days (0.00 posts per day).
20 Most recent posts:
There seems to be a lack of research regarding the use of tick-by-tick data as input to machine learning models. Has anyone experimented with machine learning and tick-by-tick data? I’ve trained an LSTM with about 2 years worth using tick-by-tick data with DTN which fit a certain criteria such as float, volume, price. The result is a model with around 69.9% accuracy. A naïve model which predicts that the bid price will be the same price as the last tick has an accuracy of around 65%. I’m wondering if I can increase my model’s accuracy through feature engineering. Could anyone share research papers regarding machine learning and tick-by-tick data? Does anyone have any insight regarding data transformations that can be applied to financial data which may result in increased accuracy if used as a feature in machine learning models?
Interests & Tools: Machine Learning, Neural Networks, Deep Learning, Python, Java, Trading, Small Caps, Interactive Brokers. Edited by keohir808 on Jan 30, 2022 at 08:50 PM
For historical data, is there a way to retrieve the high,low,open,close for each trade returned by the HTT command?
For example, when level 1 data is streamed, the Update/Message data contains data such as the high,low,open,close for each trade.
Interests & Tools: Java, Trading, Small Caps, Interactive Brokers. Edited by keohir808 on Feb 29, 2020 at 07:06 PM
Please disregard. It was an error on my part. 
Java, Trading, Small Caps, Interactive Brokers.
Hello, had a question about forming 5 minute bar charts with the BW command. Let's say I send this input:
BW,AMD,300,20190920 150000,,,,160000,,s,,1
Where:
[Interval] is 300 [BeginDate BeginTime] is 20190920 150000 [EndFilterTime] is 160000
I noticed that the output is:
BH,AMD,2019-09-20 15:10:00,29.9916,30.0100,29.9603,29.9900,40499085,186030,0, BH,AMD,2019-09-20 15:20:00,29.9910,30.0499,29.9850,30.0268,41155771,306028,0, BH,AMD,2019-09-20 15:30:00,30.0401,30.0500,30.0100,30.0300,41785831,299608,0, BH,AMD,2019-09-20 15:40:00,29.9700,29.9900,29.9600,29.9850,42484285,275326,0, BH,AMD,2019-09-20 15:50:00,29.9700,29.9750,29.8600,29.8850,43546108,641482,0, BH,AMD,2019-09-20 16:00:00,29.9900,30.0800,29.9750,30.0400,47936227,3503308,0,
The data is being sent back in increments of 10 minutes. E.g.
15:10:00, 15:20:00, 15:30:00, 15:40:00, ...
Shouldn't it instead be returning data in increments of 5 minutes since 300 was specified as the interval? E.g.
15:05:00, 15:10:00, 15:15:00, 15:20:00, 15:25:00, ... Edited by keohir808 on Sep 20, 2019 at 04:50 PM
I thought that possibly, the cause was because I was on trial, but it looks like after I subscribed to data, the most recent trade aggressor field still returns empty. Edited by keohir808 on Sep 18, 2019 at 09:08 AM
Hello, I noticed that the "most recent trade aggressor" field always seems to be empty.
The output:
echo: S,SERVER DISCONNECTED echo: S,CURRENT PROTOCOL,6.1, echo: S,CURRENT UPDATE FIELDNAMES,Symbol,Bid,Ask,Spread,Most Recent Trade,Most Recent Trade Size,Most Recent Trade Aggressor echo: S,SERVER CONNECTED echo: S,CUST,real_time,66.112.156.223,60004,V7Cm0mvJNxcYEwJ,6.1.0.20,0, TRIAL NYSE NASDAQ-L2 NASDAQ L2_SERV AMEX RT_TRADER COMTEX PRIMEZONE DTNNEWS SCROLLNEWS EXFUND BENZINGA INDEX L2 ,Z3TRL1O122Ab1X2Ag1D2DE2EF1a,500,QT_API,20190918, echo: P,YUMA,5.3600,5.3800,0.0200,5.3782,5,, echo: Q,YUMA,5.3600,5.3700,0.0100,5.3782,5,, echo: Q,YUMA,5.3500,5.3700,0.0200,5.3782,5,, echo: Q,YUMA,5.3500,5.3700,0.0200,5.3782,5,, echo: Q,YUMA,5.3500,5.3700,0.0200,5.3782,5,,
Looking at the IQFeed Connection Manager, I see IQFeed Version: 6.1.0.20
I see this field is described as, "Limited exchanges - Provided by the exchange or 3rd party data provider". Could this be the reason? I can't seem to find an exchange or a stock where the "Most Recent Trade Aggressor" field doesn't return empty. Edited by keohir808 on Sep 16, 2019 at 09:16 PM
|