class Spectrum: def __init__(self, price_step): self.price_step = price_step self.buy_bins = [] self.all_buy_bins = [] self.all_buy_volumes = [] self.all_sell_bins = [] self.all_sell_volumes = [] self.sell_bins = [] self.best_sell = -1 self.best_buy = -1 self.buy_volumes = [0 for i in range(50)] self.sell_volumes = [0 for i in range(50)] self.spectrum_sell_bins = [] self.spectrum_buy_bins = [] self.spectrum_sell = [0 for i in range(10)] self.spectrum_buy = [0 for i in range(10)] def insert_to_buy(self, price, volume): max_price = self.all_buy_bins[0] # new one is greater than the currently recorded maximum if price > max_price: dist = math.ceil((price - max_price) / self.price_step) self.all_buy_bins = [i for i in np.arange(price, max_price, -self.price_step)] + self.all_buy_bins self.all_buy_volumes = [0 for i in range(len(self.all_buy_bins) - len(self.all_buy_volumes))] + self.all_buy_volumes self.all_buy_volumes[0] += volume return 0 else: idx = math.ceil((max_price - price) / self.price_step) if idx < len(self.all_buy_bins): self.all_buy_volumes[idx] += volume return idx else: dist = idx - len(self.all_buy_bins) + 1 self.all_buy_bins = self.all_buy_bins + [i for i in np.arange(self.all_buy_bins[-1] - self.price_step, price - 1, -self.price_step)] self.all_buy_volumes = self.all_buy_volumes + [0 for i in range(len(self.all_buy_bins) - len(self.all_buy_volumes))] self.all_buy_volumes[idx] += volume return idx def insert_to_sell(self, price, volume): min_price = self.all_sell_bins[0] # new one is less than the currently recorded minimum if price < min_price: dist = math.ceil((min_price - price) / self.price_step) self.all_sell_bins = [i for i in np.arange(price, min_price, self.price_step)] + self.all_sell_bins self.all_sell_volumes = [0 for i in range(len(self.all_sell_bins) - len(self.all_sell_volumes))] + self.all_sell_volumes self.all_sell_volumes[0] += volume return 0 else: idx = math.ceil((price - min_price) / self.price_step) if idx < len(self.all_sell_bins): self.all_sell_volumes[idx] += volume return idx else: dist = idx - len(self.all_sell_bins) + 1 self.all_sell_bins = self.all_sell_bins + [i for i in np.arange(self.all_sell_bins[-1] + self.price_step, price + 1, self.price_step)] self.all_sell_volumes = self.all_sell_volumes + [0 for i in range(len(self.all_sell_bins) - len(self.all_sell_volumes))] self.all_sell_volumes[idx] += volume return idx def delete_from_buy(self, price, volume): max_price = self.all_buy_bins[0] idx = math.ceil((max_price - price) / self.price_step) if 0 <= idx < len(self.all_buy_bins): if volume < self.all_buy_volumes[idx]: self.all_buy_volumes[idx] -= volume return idx # find first non-zero element else: self.all_buy_volumes[idx] = 0 while self.all_buy_volumes[idx] == 0: if (idx == len(self.all_buy_volumes) - 1): break idx += 1 return idx else: return -1 def delete_from_sell(self, price, volume): min_price = self.all_sell_bins[0] idx = math.ceil((price - min_price) / self.price_step) if 0 <= idx < len(self.all_sell_bins): if volume < self.all_sell_volumes[idx]: self.all_sell_volumes[idx] -= volume return idx # find first non-zero element else: self.all_sell_volumes[idx] = 0 while self.all_sell_volumes[idx] == 0: if (idx == len(self.all_sell_volumes) - 1): break idx += 1 return idx else: return -1 def find_idx_sell(self, price): k = math.ceil((price - self.best_sell) / self.price_step) if k == 50: k = 49 return int(k) def find_idx_buy(self, price): k = math.ceil((self.best_buy - price) / self.price_step) if k == 50: k = 49 return int(k) def find_idx_spectrum_sell(self, price): k = math.ceil((price - self.best_sell) / self.price_step) // 5 if k == 10: k = 9 return k def find_idx_spectrum_buy(self, price): k = math.ceil((self.best_buy - price) / self.price_step) // 5 if k == 10: k = 9 return k def recalc_spectrum_sell(self): self.spectrum_sell_bins = [self.sell_bins[i] for i in range(0, 50, 5)] self.spectrum_sell = [sum(self.sell_volumes[i:i+5]) for i in range(0, 50, 5)] def recalc_spectrum_buy(self): self.spectrum_buy_bins = [self.buy_bins[i] for i in range(0, 50, 5)] self.spectrum_buy = [sum(self.buy_volumes[i:i+5]) for i in range(0, 50, 5)] def new_sell_order(self, price, volume): # no sell orders recorded yet if self.best_sell == -1: self.best_sell = price max_sell = self.best_sell + 50 * self.price_step self.sell_bins = [p for p in np.arange(self.best_sell, max_sell, self.price_step)] self.spectrum_sell_bins = [p for p in np.arange(self.best_sell, max_sell, self.price_step * 5)] self.sell_volumes[0] = volume self.spectrum_sell[0] = volume self.all_sell_bins = self.sell_bins.copy() self.all_sell_volumes = self.sell_volumes.copy() else: # sell order falls somewhere in the existing bins if self.best_sell <= price < self.best_sell + 50 * self.price_step: idx = self.find_idx_sell(price) if idx == 50: idx = 49 self.sell_volumes[idx] += volume spect_idx = self.find_idx_spectrum_sell(price) self.spectrum_sell[spect_idx] += volume _ = self.insert_to_sell(price, volume) else: # found new best, update everything if self.best_sell > price: idx = self.insert_to_sell(price, volume) self.best_sell = price if idx + 50 < len(self.all_sell_bins): self.sell_bins = self.all_sell_bins[idx:idx+50] self.sell_volumes = self.all_sell_volumes[idx:idx+50] else: self.sell_bins = [p for p in np.arange(self.best_sell, self.best_sell + 50 * self.price_step, self.price_step)] self.sell_volumes = self.all_sell_volumes[idx:] + [0 for i in range(50 - len(self.all_sell_volumes) + idx)] self.recalc_spectrum_sell() # save for the later usage else: _ = self.insert_to_sell(price, volume) def new_buy_order(self, price, volume): # no buy orders recorded yet if self.best_buy == -1: self.best_buy = price min_buy = self.best_buy - 50 * self.price_step self.buy_bins = [p for p in np.arange(self.best_buy, min_buy, -self.price_step)] self.spectrum_buy_bins = [p for p in np.arange(self.best_buy, min_buy, -self.price_step * 5)] self.buy_volumes[0] = volume self.spectrum_buy[0] = volume self.all_buy_bins = self.buy_bins.copy() self.all_buy_volumes = self.buy_volumes.copy() else: # buy order falls somewhere in the existing bins if self.best_buy >= price > self.best_buy - 50 * self.price_step: idx = self.find_idx_buy(price) if idx == 50: idx = 49 self.buy_volumes[idx] += volume spect_idx = self.find_idx_spectrum_buy(price) self.spectrum_buy[spect_idx] += volume _ = self.insert_to_buy(price, volume) else: # found new best, update everything if self.best_buy < price: idx = self.insert_to_buy(price, volume) self.best_buy = price if idx + 50 < len(self.all_buy_bins): self.buy_bins = self.all_buy_bins[idx:idx+50] self.buy_volumes = self.all_buy_volumes[idx:idx+50] else: self.buy_bins = [p for p in np.arange(self.best_buy, self.best_buy - 50 * self.price_step, -self.price_step)] self.buy_volumes = self.all_buy_volumes[idx:] + [0 for i in range(50 - len(self.all_buy_volumes) + idx)] self.recalc_spectrum_buy() # save for the later usage else: _ = self.insert_to_buy(price, volume) def delete_sell_order(self, price, volume): # does not remove current best if self.best_sell + 50 * self.price_step > price > self.best_sell or price == self.best_sell and volume < self.sell_volumes[0]: idx = self.find_idx_sell(price) self.sell_volumes[idx] = max(0, self.sell_volumes[idx] - volume) spect_idx = self.find_idx_spectrum_sell(price) self.spectrum_sell[spect_idx] = max(0, self.spectrum_sell[spect_idx] - volume) else: # if removes current best if price == self.best_sell and volume >= self.sell_volumes[0]: idx = self.delete_from_sell(price, volume) self.best_sell = self.all_sell_bins[idx] if idx + 50 < len(self.all_sell_bins): self.sell_bins = self.all_sell_bins[idx:idx+50] self.sell_volumes = self.all_sell_volumes[idx:idx+50] else: self.sell_bins = [p for p in np.arange(self.best_sell, self.best_sell + 50 * self.price_step, self.price_step)] self.sell_volumes = self.all_sell_volumes[idx:] + [0 for i in range(50 - len(self.all_sell_volumes) + idx)] self.recalc_spectrum_sell() # if does not fall in 50 steps elif price > self.best_sell + 50 * self.price_step: _ = self.delete_from_sell(price, volume) def delete_buy_order(self, price, volume): # does not remove current best if self.best_buy - 50 * self.price_step < price < self.best_buy or price == self.best_buy and volume < self.buy_volumes[0]: idx = self.find_idx_buy(price) self.buy_volumes[idx] = max(0, self.buy_volumes[idx] - volume) spect_idx = self.find_idx_spectrum_buy(price) self.spectrum_buy[spect_idx] = max(0, self.spectrum_buy[spect_idx] - volume) else: # if removes current best if price == self.best_buy and volume >= self.buy_volumes[0]: idx = self.delete_from_buy(price, volume) self.best_buy = self.all_buy_bins[idx] if idx + 50 < len(self.all_buy_bins): self.buy_bins = self.all_buy_bins[idx:idx+50] self.buy_volumes = self.all_buy_volumes[idx:idx+50] else: self.buy_bins = [p for p in np.arange(self.best_buy, self.best_buy - 50 * self.price_step, -self.price_step)] self.buy_volumes = self.all_buy_volumes[idx:] + [0 for i in range(50 - len(self.all_buy_volumes) + idx)] self.recalc_spectrum_buy() # if does not fall in 50 steps elif price > self.best_buy + 50 * self.price_step: _ = self.delete_from_buy(price, volume)