Each trader sees real-time bar charts that are unique to them

Discussion in 'Automated Trading' started by abattia, Mar 23, 2010.

  1. Please bear with me as I wade through the following (I’m not claiming anything original here; just pulling together elements proposed by other contributors in other threads to see if it all holds together) ...

    1) Each of us uses a different trading computer.

    2) Each trading computer constructs bars from the ticks that arrive at that computer, referenced against that computer’s internal clock. When the computer judges each bar’s time period has passed, it “closes” that bar (which determines that bar’s OHLC), and begins to count subsequent ticks towards the next bar.

    3) As A) we all likely have our computers’ internal clocks synchronized slightly differently, and B) we each have our trading computer located a different “Internet distance” from our source of tick data, and C) we each source our tick data from a variety of combinations of sources anyway (e.g. from brokers, from eSignal, from exchanges, etc.)...

    ... the precise combination of TICKS and WHEN THEY ARRIVE (as referenced by each computer’s internal clock) is going to be unique to each one of us.

    Therefore the precise combination of OHLCs across all the bars in a given real-time chart is also going to be different for each one of us (with bar Open and Close perhaps the most volatile elements). WE WILL EACH SEE A DIFFERENT REAL-TIME CHART; EVEN FOR THE SAME INSTRUMENT AND TIME INTERVAL.

    So where does this lead? (And thanks for bothering to read this far... nearly there!)

    In designing a mechanical intraday trading strategy, one could make use of entry (or exit) signals sensitive to specific sequences of intraday bar-to-bar OHLC measurements (e.g. “after a lower low followed by three successive higher closes, go long” ... or more subtly, “after MACD diff crosses above 0, go long” ... or “after 12-period momentum goes below 0, exit position” ... etc).

    However, given that the real-time OHLC sequences perceived by each of us are going to be subject to our specific situation (computer clock, “internet distance”, and data source, and perhaps other elements, too), WE WON’T BE ABLE TO DRAW ANY MEANINGFUL CONCLUSIONS ABOUT THE VALIDITY OF OUR STRATEGY THROUGH BACKTESTING (OR FOWARDTESTING) UNLESS WE ARE ABLE TO KEEP ALL ELEMENTS OF OUR SPECIFIC SITUATION THE SAME FOR BOTH “LIVE” TRADING AND TESTING.

    Also, even though a strategy including these subjective OHLC sensitive entry/exit signals might work for one trader in their specific situation, exactly the same strategy might well not be portable to another trader (if they have clock synced differently, are in a different location, and use a different data source) or even to another computer.


    Superior entry/exit signals (that have a more objective/absolute character from an intraday perspective) can instead be looked for among the likes of, for example, daily pivots (as these depend on daily - not intraday - HLC, which there can be much less argument about, and so all traders see the same), or fading extreme moves of ^TICK, etc.

    Any thoughts? Comments?
  2. Yes, I have seen this occurrence.

    For example take the

    1) Transact

    2) OEC

    3) Barchart.com

    Live data feeds for ES MINI and put them on a 233 tick chart.

    I believe they are slightly different because of the way GLOBEX reports orders. But from the same data provider they will look the same to all other participants I believe.

    Using VOLUME BASED CHARTS I heard is the most definitive way to be accurate. It again has something to do with trades vs. volume and how GLOBEX reports. Anyone can feel free to correct me.
  3. thstart


    Just get the data from where you will trade.
  4. Thank you.

    Yes, I can see that a TICK CHART would get round both the computer clock syncing, and "Internet distance" issues. I have not made use of tick charts, so must look further to see whether OHLC sensitive signals might work better with these. Thank you!

    VOLUME BASED CHARTS I have not looked into either. Something else I'd better find out more about.
  5. Yes, like 6000v chart should look the same to all participants or almost exact due to the way GLOBEX reports. That is only GLOBEX though, so that doesn't cover everything..

    Again anyone can correct me where I'm wrong
  6. The challenge with volume or tick charts is how far back does your chart go, and at what time of the day did it start at? If two traders started their charts at different times, the close of each bar will be at a different time.

    Not that I personally care cause I don't trade that way, but thought I'd point it out.
  7. Bingo. You are on the right track! Nothing makes less sense than trading from a chart!!
  8. I discovered candlestick charts were significantly different somewhere 6 months after I started trading. That's the reason I stopped trading an exact candlestick strategy from a chart, and very much if it's a 1-min chart. I do still think candlestick analysis can be valid on higher timeframes like 15 min, but something like trading pin bars on 1 min is totally stupid imo

    I just have one thought in mind: trade the trend, or trade countertrend with lighter size (assuming countertrend is more dangerous). ofcourse trend is still in the eye of the beholder.
  9. thstart


    Candlestick bars do not make sense to be used intra-day.
  10. jprad


    Ticks contain a timestamp that's created at the source. If your trading platform is ignoring that timestamp and using the internal clock to build bars then it is incredibly broken.
    #10     Mar 23, 2010