thatsjonsense / ticktock Goto Github PK

Google offers historical prices at https://www.google.com/finance/getprices?i=60&p=1d&f=d,o,h,l,c,v&df=cpct&q=IBM more info and alternatives here http://www.quantshare.com/sa-426-6-ways-to-download-free-intraday-and-tick-data-for-the-us-stock-market

Plan is: get 1-second level data over a day for ~100 stocks, store in a CSV. Write the meteor server code to parse that CSV, load it into some kind of model, then spit out data from there in place of real stock market connections.

Automatically do that any time market is closed

Right now we store quotes for a stock -- meaning it's price at a given date. Asking for the price of a stock at a certain time just means finding the most recent quote before that moment.

A good quote is fresh -- it comes from within seconds or minutes of the moment you asked for. A bad quote is stale -- we should either say the price is null at that point (because we have no clue), or immediately grab a fresher quote.

So how stale is too stale? A couple challenges come in

• On weekends, we don't get a quote for 2 days, but the latest price is still right
• For recent data, we can get up to the second quotes. For the distant past, it's only daily.

Right now we're returning quotes while the market is closed, even though the last trade is from hours before.

The "time" in a quote should be the time the last trade happened. And we shouldn't store quotes while the market is closed: that just clogs things up.

So we need to parse the LastTradeTime coming back from Y! (let's assume all EST for now) and store that as a quote time. And, we need to avoid logging duplicate quotes in that case.

Eg LNKD will start trading at 9:32am, finish at 4:06pm. We'd expect the window to be 9:30 to 4 exactly. Are we parsing the times wrong or is Google's data funky?