# Last edited on 2014-02-27 22:26:15 by stolfilocal [b]Checking the Chinese Slumber Method predictions for Feb/27[/b] Prediction posted on: Thursday 2014-02-27, 04:04 UTC Prediction was valid for: Thursday 2014-02-27, 19:00--19:59 UTC Recall that I had given two predictions, one for bulls, one for bears. The actual price was between the two, closer to the bullish one: Huobi's bearish predicted price: 2781 CNY Huobi's bullish predicted price: 3802 CNY Huobi's actual price (L+H)/2: 3533 CNY Error of bullish prediction: 269 CNY Bitstamp's bearish predicted price: 454 USD Bitstamp's bullish predicted price: 621 USD Bitstamp's actual price (L+H)/2: 585 USD Error of bullish prediction: 36 USD The price ratio between Huobi and Bitstamp had its normal value R ~ 6.12. It seems that arbitrage trading has been restored. The [b]Huobi[/b] prediction in question is the rightmost blue rectangle on the chart below. The light blue-gray rectangles are the previous predictions. The orange and grey dots are the True and False Slumber Points, the mean Huobi prices at 19:00 UTC every day. The orange line is the trend that was assumed for the prediction. [url=http://www.ic.unicamp.br/~stolfi/temp/2014-02-27-prediction-check-HUBI-annotated.png][img width=600]http://www.ic.unicamp.br/~stolfi/temp/2014-02-27-prediction-check-HUBI-annotated.png[/img][/url] The following chart shows the [b]Bitstamp[/b] prices and predictions. The orange and grey dots are [b]Huobi[/b]'s prices at 19:00 UTC every day, divided by R = 6.12. The orange line is the trend used in the prediction, which is Huobi's trend divided by R. [url=http://www.ic.unicamp.br/~stolfi/temp/2014-02-27-prediction-check-BSTP-annotated.png][img width=600]http://www.ic.unicamp.br/~stolfi/temp/2014-02-27-prediction-check-BSTP-annotated.png[/img][/url] ---------------------------------------------------------------------- [b]Chinese Slumber Method prediction for Friday Feb/28[/b] Trade volume for today Feb/27 volume (~123 kBTC) was about half of yesterday's (~251 kBTC), and apparently broke the seven-day spell of hectic trading at Huobi where the price, even at Slumber times, seemed to have no definite trend. It is as if the seven days Feb/20--26 had been one single "trading session", like Feb/11--12, Feb/13--14, and Feb/16--17. The price at the official Slumber hour (19:00--19:59) hardly changed from yesterday's: from 3543 CNY to 3533 CNY. Eeven though today was nominally a False Slumber Point, it suggests a rather simple pattern that seems to fit most of the True Slumber points since Feb/06. The general form of the trend is a shifted decaying exponential, p = A + B*Q**d, where A,B,Q are constants and d is the day of the month. The ratio Q, fitted by hand computations, is 0.66 (i.e. a 34% decay for each day of the part of the price above the constant A). Since Feb/20 the term B*Q**d is so small that the trend can be approximated by a horizontal line p = A. That trend is broken into three pieces by two sharp and permanent down-steps, one by ~232 CNY on Feb/10 (the first press release by MtGO), and the other by ~372 CNY on Feb/20 (the second press release). Thus each piece has different values of A and B, which can be found by least squares fitting on the True Slumber points. See the figure: [url=http://www.ic.unicamp.br/~stolfi/temp/2014-02-27-prediction-make-HUBI-annotated.png][img width=600]http://www.ic.unicamp.br/~stolfi/temp/2014-02-27-prediction-make-HUBI-annotated.png[/img][/url] The prediction for today (blue rectangle) is obtained by extrapolating the last part of this stepped-exponential trend line --- that is, repeating today's value: [b]Prediction valid for:[/b] Friday 2014-02-28, 19:00--19:59 UTC (not before, not after) [b]Huobi's predicted price:[/b] 3553 CNY. [b]Bitstamp's predicted price:[/b] 577 USD. For Bitstamp, as usual, I used Huobi's prediction divided by R = 6.12. [b]NOTE[/b]: As you can see, predicting the past is much easier than predicting the future.