Intraday Historic data reliability issues

coolkoti
Hi,
The historic data API doesn't seem to provide reliable data in many cases. One example is given below, but I can provide several more like this one. Can you please suggest if this API is usable at all? If not, what is it charged for?

Exchange: NSE
Symbol: AJANTPHARM
Dates: 2015-Mar-18 to 2015-Mar-20
Issue - The price suddenly jumps to 3000+ from 1200 range for 19-Mar and comes back on 20-Mar. Screenshot attached.

This is still glaringly noticeable. The main problem is, there are several instances where the differences are substantial to cause serious issues in back test results, but very difficult to identify that the problem is due to the bad quality of data.

Can someone seriously look into this issue and refresh all old data if possible? (I'm sure you have the NSE data feed with which you can probably fix this once and for all?)


  • coolkoti
    Here is the list of some more discrepancies I found just by scanning for the price difference of 5% or more between EOD & intraday data (Difference between EOD high or low to the high/low price of any interval 1/5/15min):

    BPCL - 2018-10-05 (EOD H/L 298.05/238.55, Intraday H/L 281.5/239.4)
    HINDPETRO - 2018-10-05 (EOD H/L 195.45/163.0, Intraday H/L 184.6/163.0)
    PFC - 2018-10-25 (EOD H/L 85.95/76.1, Intraday H/L 85.35/80.85)
    YESBANK - 2018-09-21 (EOD H/L 287.3/218.1, Intraday H/L 271.35/222.0)
    KTKBANK - 2018-07-25 (EOD H/L 112.65/101.8, Intraday H/L 105.9/101.8)
    PCJEWELLER - 2018-07-16 (EOD H/L 107.95/83.95 , Intraday H/L 101.95 /84.25)
    TATACOMM - 2018-10-11 (EOD H/L 497.75/445.5, Intraday H/L 497.75/469.1)
    SREINFRA - 2018-11-01 (EOD H/L 36.2/30.6 , Intraday H/L 36.2/32.85)

    I did this only for year 2018 and only for the above stocks, because I got some silly results in my backtest which prompted to check data. I don't know how many more discrepancies out there. I know, there can be small differences compared to EOD data, but more than 5% is plainly bad data. The fact that there are so many discrepancies really concerns me as we all do backtest against this data and I'm not sure if we can rely on this anymore. This must be corrected immediately or remove the historic data API altogether, so that traders don't get mislead and lose money.
  • coolkoti
    Hi, Can yo update on this?
  • sujith
    Can you let us know what are you comparing here? Your query is not clear.
  • sujith
    Are you comparing day candle data to first-minute candle of the day?
  • coolkoti
    No. I'm comparing the day candle High & Low with intraday 1 min high and low (Max High & Min Low of all the 1min candles in a day). They both should match. I can understand there may be small differences in Open & Close when compared to day candle, but High & Low should ideally be the same. Even if there is a small difference of less than 0.5% from the day candle is tolerable, but the data I mentioned has more than 5% difference !!
    For example, on 2018-10-05, BPCL day candle High is 298.05 and Low is 238.55, But if I look at all 1min candles of that day and compute High is 281.5 and Low is 239.4.

    The case of AJANTPHARM is clearly an issue with the adjustment logic. For some reason, the prices were adjusted for all days except for 19-Mar.
  • rakeshr
    @coolkoti
    I'm comparing the day candle High & Low with intraday 1 min high and low (Max High & Min Low of all the 1min candles in a day). They both should match.
    No, day's OHLC and intra-day OHLC might not match always. You can go through this thread to understand more about it.
  • coolkoti
    OK. I've read the thread. There are 2 issues here:
    1. Daily & Intraday OHLC not matching - I do understand there can be small differences, but more than 5% difference?? This causes issues especially in automated trading. I think, I may need to update the OHLC of the historic 1 min data with the OHLC of the daily data. Updating Open & Close is easy, High/Low will be a bit tricky.
    2. The issue with AJANTPHARM on 19-Mar-15 mentioned above is definitely not because of the missing ticks. It must be due to the adjustments not applied correctly. I also see this issue with BATAINDIA between 05 to 07 Oct 15. Can you please correct this?
    Thanks for support.
  • rakeshr
    @coolkoti
    I also see this issue with BATAINDIA between 05 to 07 Oct 15. Can you please correct this?
    Yes, it will be adjusted according to the required stock split ratio.
Sign In or Register to comment.