Class: SQA::StrategyGenerator

Inherits:
Object
  • Object
show all
Defined in:
lib/sqa/strategy_generator.rb

Defined Under Namespace

Classes: Pattern, PatternContext, ProfitablePoint

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(stock:, min_gain_percent: 10.0, min_loss_percent: nil, fpop: 10, inflection_window: 3, max_fpl_risk: nil, required_fpl_directions: nil) ⇒ StrategyGenerator

Returns a new instance of StrategyGenerator.



144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
 
# File 'lib/sqa/strategy_generator.rb', line 144

def initialize(stock:, min_gain_percent: 10.0, min_loss_percent: nil, fpop: 10, inflection_window: 3, max_fpl_risk: nil, required_fpl_directions: nil)
  @stock = stock
  @min_gain_percent = min_gain_percent
  @min_loss_percent = min_loss_percent || -min_gain_percent  # Symmetric loss threshold
  @fpop = fpop  # Future Period of Performance
  @inflection_window = inflection_window  # Window for detecting local min/max
  @max_fpl_risk = max_fpl_risk  # Optional: Filter by max acceptable risk (volatility)
  @required_fpl_directions = required_fpl_directions  # Optional: [:UP, :DOWN, :UNCERTAIN, :FLAT]
  @profitable_points = []
  @patterns = []

  # Configure which indicators to analyze
  @indicators_config = {
    rsi: { period: 14, oversold: 30, overbought: 70 },
    macd: { fast: 12, slow: 26, signal: 9 },
    stoch: { k_period: 14, d_period: 3, oversold: 20, overbought: 80 },
    sma_cross: { short: 20, long: 50 },
    ema: { period: 20 },
    bbands: { period: 20, nbdev: 2.0 },
    volume: { period: 20, threshold: 1.5 }
  }
end

Instance Attribute Details

#fpop ⇒ Object (readonly)

Returns the value of attribute fpop.



140
141
142
 
# File 'lib/sqa/strategy_generator.rb', line 140

def fpop
  @fpop
end

#indicators_config ⇒ Object (readonly)

Returns the value of attribute indicators_config.



140
141
142
 
# File 'lib/sqa/strategy_generator.rb', line 140

def indicators_config
  @indicators_config
end

#inflection_window ⇒ Object (readonly)

Returns the value of attribute inflection_window.



140
141
142
 
# File 'lib/sqa/strategy_generator.rb', line 140

def inflection_window
  @inflection_window
end

#max_fpl_risk ⇒ Object (readonly)

Returns the value of attribute max_fpl_risk.



140
141
142
 
# File 'lib/sqa/strategy_generator.rb', line 140

def max_fpl_risk
  @max_fpl_risk
end

#min_gain_percent ⇒ Object (readonly)

Returns the value of attribute min_gain_percent.



140
141
142
 
# File 'lib/sqa/strategy_generator.rb', line 140

def min_gain_percent
  @min_gain_percent
end

#min_loss_percent ⇒ Object (readonly)

Returns the value of attribute min_loss_percent.



140
141
142
 
# File 'lib/sqa/strategy_generator.rb', line 140

def min_loss_percent
  @min_loss_percent
end

#patterns ⇒ Object (readonly)

Returns the value of attribute patterns.



140
141
142
 
# File 'lib/sqa/strategy_generator.rb', line 140

def patterns
  @patterns
end

#profitable_points ⇒ Object (readonly)

Returns the value of attribute profitable_points.



140
141
142
 
# File 'lib/sqa/strategy_generator.rb', line 140

def profitable_points
  @profitable_points
end

#required_fpl_directions ⇒ Object (readonly)

Returns the value of attribute required_fpl_directions.



140
141
142
 
# File 'lib/sqa/strategy_generator.rb', line 140

def required_fpl_directions
  @required_fpl_directions
end

#stock ⇒ Object (readonly)

Returns the value of attribute stock.



140
141
142
 
# File 'lib/sqa/strategy_generator.rb', line 140

def stock
  @stock
end

Instance Method Details

#discover_context_aware_patterns(analyze_regime: true, analyze_seasonal: true, sector: nil) ⇒ Array<Pattern>

Discover patterns with context (regime, seasonal, sector)

Parameters:

  • analyze_regime (Boolean) (defaults to: true) —

    Detect and filter by market regime

  • analyze_seasonal (Boolean) (defaults to: true) —

    Detect seasonal patterns

  • sector (Symbol) (defaults to: nil) —

    Sector classification

Returns:

  • (Array<Pattern>) —

    Patterns with context metadata



379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
 
# File 'lib/sqa/strategy_generator.rb', line 379

def discover_context_aware_patterns(analyze_regime: true, analyze_seasonal: true, sector: nil)
  puts "\n" + "=" * 70
  puts "Context-Aware Pattern Discovery"
  puts "=" * 70

  # Step 1: Detect market regime
  if analyze_regime
    regime_data = SQA::MarketRegime.detect(@stock)
    puts "Current regime: #{regime_data[:type]} (#{regime_data[:strength]} strength)"

    # Split data by regime
    regime_splits = SQA::MarketRegime.split_by_regime(@stock)

    puts "\nRegime periods:"
    regime_splits.each do |regime, periods|
      total_days = periods.sum { |p| p[:duration] }
      puts "  #{regime}: #{total_days} days across #{periods.size} periods"
    end
  end

  # Step 2: Analyze seasonality
  if analyze_seasonal
    seasonal_data = SQA::SeasonalAnalyzer.analyze(@stock)
    puts "\nSeasonal analysis:"
    puts "  Best months: #{seasonal_data[:best_months].join(', ')}"
    puts "  Worst months: #{seasonal_data[:worst_months].join(', ')}"
    puts "  Best quarters: Q#{seasonal_data[:best_quarters].join(', Q')}"
    puts "  Has seasonal pattern: #{seasonal_data[:has_seasonal_pattern]}"
  end

  # Step 3: Discover patterns normally
  patterns = discover_patterns

  # Step 4: Add context to each pattern
  patterns.each do |pattern|
    if analyze_regime
      pattern.context.market_regime = regime_data[:type]
      pattern.context.volatility_regime = regime_data[:volatility]
    end

    if analyze_seasonal && seasonal_data[:has_seasonal_pattern]
      pattern.context.valid_months = seasonal_data[:best_months]
      pattern.context.valid_quarters = seasonal_data[:best_quarters]
    end

    if sector
      pattern.context.sector = sector
    end

    # Add discovery period

    date_column = @stock.df.data.columns.include?("date") ? "date" : "timestamp"
    dates = @stock.df[date_column].to_a

    pattern.context.discovered_period = "#{dates.first} to #{dates.last}"
  end

  puts "\n" + "=" * 70
  puts "Context-Aware Discovery Complete"
  puts "  Patterns found: #{patterns.size}"
  puts "  Patterns with context: #{patterns.count { |p| p.context.valid? }}"
  puts "=" * 70

  patterns
end

#discover_patterns(min_pattern_frequency: 2) ⇒ Object

Main entry point: Discover patterns in historical data



168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
 
# File 'lib/sqa/strategy_generator.rb', line 168

def discover_patterns(min_pattern_frequency: 2)
  puts "=" * 70
  puts "Strategy Generator: Discovering Profitable Patterns"
  puts "=" * 70
  puts "Target gain: ≥#{min_gain_percent}%"
  puts "Target loss: ≤#{min_loss_percent}%"
  puts "FPOP (Future Period of Performance): #{fpop} days"
  puts "Inflection window: #{inflection_window} days"
  puts

  # Step 1: Find profitable inflection points
  find_profitable_points

  return [] if @profitable_points.empty?

  # Step 2: Calculate indicators at each profitable point
  analyze_indicator_states

  # Step 3: Mine patterns from indicator states
  mine_patterns(min_frequency: min_pattern_frequency)

  # Step 4: Calculate pattern statistics
  calculate_pattern_statistics

  @patterns
end

#export_patterns(filename) ⇒ Object

Export patterns to CSV



241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
 
# File 'lib/sqa/strategy_generator.rb', line 241

def export_patterns(filename)
  require 'csv'

  CSV.open(filename, 'w') do |csv|
    csv << ['Pattern', 'Frequency', 'Avg Gain %', 'Avg Holding Days', 'Success Rate %', 'Conditions']

    @patterns.each_with_index do |pattern, i|
      conditions_str = pattern.conditions.map { |k, v| "#{k}=#{v}" }.join('; ')
      csv << [
        i + 1,
        pattern.frequency,
        pattern.avg_gain.round(2),
        pattern.avg_holding_days.round(1),
        pattern.success_rate.round(2),
        conditions_str
      ]
    end
  end

  puts "Patterns exported to #{filename}"
end

#generate_strategies(top_n: 5, strategy_type: :class) ⇒ Object

Generate multiple strategies from top N patterns



214
215
216
217
218
 
# File 'lib/sqa/strategy_generator.rb', line 214

def generate_strategies(top_n: 5, strategy_type: :class)
  @patterns.take(top_n).map.with_index do |pattern, i|
    generate_strategy(pattern_index: i, strategy_type: strategy_type)
  end
end

#generate_strategy(pattern_index: 0, strategy_type: :proc) ⇒ Object

Generate a trading strategy from discovered patterns



196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
 
# File 'lib/sqa/strategy_generator.rb', line 196

def generate_strategy(pattern_index: 0, strategy_type: :proc)
  return nil if @patterns.empty?

  pattern = @patterns[pattern_index]

  case strategy_type
  when :proc
    generate_proc_strategy(pattern)
  when :class
    generate_class_strategy(pattern)
  when :kbs
    generate_kbs_strategy(pattern)
  else
    raise "Unknown strategy type: #{strategy_type}"
  end
end

#print_patterns(max_patterns: 10) ⇒ Object

Print discovered patterns



221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
 
# File 'lib/sqa/strategy_generator.rb', line 221

def print_patterns(max_patterns: 10)
  puts "\n" + "=" * 70
  puts "Discovered Patterns (Top #{[max_patterns, @patterns.size].min})"
  puts "=" * 70

  @patterns.take(max_patterns).each_with_index do |pattern, i|
    puts "\nPattern ##{i + 1}:"
    puts "  Frequency: #{pattern.frequency} occurrences"
    puts "  Average Gain: #{pattern.avg_gain.round(2)}%"
    puts "  Average Holding: #{pattern.avg_holding_days.round(1)} days"
    puts "  Success Rate: #{pattern.success_rate.round(2)}%"
    puts "  Conditions:"
    pattern.conditions.each do |indicator, state|
      puts "    - #{indicator}: #{state}"
    end
  end
  puts
end

#walk_forward_validate(train_size: 250, test_size: 60, step_size: 30) ⇒ Hash

Walk-forward validation - discover patterns with time-series cross-validation

Splits data into train/test windows and rolls forward through history to prevent overfitting. Only keeps patterns that work out-of-sample.

Parameters:

  • train_size (Integer) (defaults to: 250) —

    Training window size in days

  • test_size (Integer) (defaults to: 60) —

    Testing window size in days

  • step_size (Integer) (defaults to: 30) —

    How many days to step forward each iteration

Returns:

  • (Hash) —

    Validation results with patterns and performance



273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
 
# File 'lib/sqa/strategy_generator.rb', line 273

def walk_forward_validate(train_size: 250, test_size: 60, step_size: 30)
  puts "\n" + "=" * 70
  puts "Walk-Forward Validation"
  puts "=" * 70
  puts "Training window: #{train_size} days"
  puts "Testing window: #{test_size} days"
  puts "Step size: #{step_size} days"
  puts

  prices = @stock.df["adj_close_price"].to_a


  date_column = @stock.df.data.columns.include?("date") ? "date" : "timestamp"
  dates = @stock.df[date_column].to_a.map { |d| Date.parse(d.to_s) }

  validated_patterns = []
  validation_results = []

  start_idx = 0
  iteration = 0

  while start_idx + train_size + test_size < prices.size
    iteration += 1
    train_start = start_idx
    train_end = start_idx + train_size
    test_start = train_end
    test_end = test_start + test_size

    puts "\nIteration #{iteration}:"
    puts "  Train: #{dates[train_start]} to #{dates[train_end - 1]}"
    puts "  Test:  #{dates[test_start]} to #{dates[test_end - 1]}"

    # Create temporary stock with training data
    train_data = create_stock_subset(train_start, train_end)

    # Discover patterns on training data
    temp_generator = SQA::StrategyGenerator.new(
      stock: train_data,
      min_gain_percent: @min_gain_percent,
      fpop: @fpop,
      inflection_window: @inflection_window,
      max_fpl_risk: @max_fpl_risk,
      required_fpl_directions: @required_fpl_directions
    )

    train_patterns = temp_generator.discover_patterns(min_pattern_frequency: 2)

    # Test each pattern on out-of-sample data
    test_data = create_stock_subset(test_start, test_end)

    train_patterns.each do |pattern|
      # Generate strategy from pattern
      strategy = temp_generator.generate_strategy(
        pattern_index: train_patterns.index(pattern),
        strategy_type: :proc
      )

      # Backtest on test period
      begin
        backtest = SQA::Backtest.new(stock: test_data, strategy: strategy)
        results = backtest.run

        # Store validation result
        validation_results << {
          iteration: iteration,
          pattern: pattern,
          train_period: "#{dates[train_start]} to #{dates[train_end - 1]}",
          test_period: "#{dates[test_start]} to #{dates[test_end - 1]}",
          test_return: results.total_return,
          test_sharpe: results.sharpe_ratio,
          test_max_drawdown: results.max_drawdown
        }

        # Keep pattern if it performed well out-of-sample
        if results.total_return > 0 && results.sharpe_ratio > 0.5
          validated_patterns << pattern
        end
      rescue => e
        puts "    Warning: Pattern validation failed: #{e.message}"
      end
    end

    start_idx += step_size
  end

  puts "\n" + "=" * 70
  puts "Validation Complete"
  puts "  Total iterations: #{iteration}"
  puts "  Total patterns tested: #{validation_results.size}"
  puts "  Patterns validated: #{validated_patterns.size}"
  puts "=" * 70

  {
    validated_patterns: validated_patterns,
    validation_results: validation_results,
    total_iterations: iteration
  }
end