ferric-alpha 0.1.0

alphalens-style high-performance factor analysis built with Rust and Polars
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
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
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
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
371
372
373
374
375
376
377
378
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
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
use std::collections::{BTreeMap, BTreeSet, HashSet};

use polars::prelude::{
    BooleanChunked, DataFrame, DataType, NamedFrom, NewChunkedArray, Series, SortMultipleOptions,
};

use crate::Period;
use crate::data::required_column;
use crate::error::{FerricAlphaError, Result};
use crate::performance::frame::{date_column, finish_frame};
use crate::performance::{PortfolioOptions, ReturnOptions};
use crate::{factor_returns, factor_weights};

#[derive(Debug)]
struct ActiveSleeve {
    expiry_index: usize,
    weights: BTreeMap<String, f64>,
}

pub fn cumulative_returns(returns: &DataFrame) -> Result<DataFrame> {
    let date = required_column(returns, "date")?;
    let values = required_column(returns, "return")?;
    if !matches!(date.dtype(), DataType::Datetime(_, _)) {
        return Err(FerricAlphaError::InvalidDtype {
            column: "date",
            expected: "Datetime",
            actual: date.dtype().to_string(),
        });
    }
    if !matches!(values.dtype(), DataType::Float64) {
        return Err(FerricAlphaError::InvalidDtype {
            column: "return",
            expected: "Float64",
            actual: values.dtype().to_string(),
        });
    }
    let has_period = returns.column("period").is_ok();
    if has_period && !matches!(returns.column("period")?.dtype(), DataType::String) {
        return Err(FerricAlphaError::InvalidDtype {
            column: "period",
            expected: "String",
            actual: returns.column("period")?.dtype().to_string(),
        });
    }
    if date.null_count() > 0 {
        return Err(FerricAlphaError::NullKey);
    }

    let sort = if has_period {
        vec!["period", "date"]
    } else {
        vec!["date"]
    };
    let sorted = returns.sort(sort.clone(), SortMultipleOptions::default())?;
    let dates = sorted.column("date")?.as_materialized_series().datetime()?;
    let values = sorted.column("return")?.as_materialized_series().f64()?;
    let periods = sorted
        .column("period")
        .ok()
        .map(|column| column.as_materialized_series().str())
        .transpose()?;

    let mut seen = HashSet::with_capacity(sorted.height());
    let mut output_dates = Vec::with_capacity(sorted.height());
    let mut output_periods = Vec::with_capacity(sorted.height());
    let mut output_wealth = Vec::with_capacity(sorted.height());
    let mut active_period: Option<String> = None;
    let mut wealth = 1.0;

    for index in 0..sorted.height() {
        let date = dates
            .physical()
            .get(index)
            .ok_or(FerricAlphaError::NullKey)?;
        let period = periods
            .as_ref()
            .map(|periods| {
                periods
                    .get(index)
                    .map(ToOwned::to_owned)
                    .ok_or(FerricAlphaError::InvalidInput {
                        operation: "cumulative_returns",
                        reason: "period must not contain null values",
                    })
            })
            .transpose()?;
        if let Some(period) = &period {
            validate_period_label(period)?;
        }
        if !seen.insert((date, period.clone())) {
            return Err(FerricAlphaError::InvalidInput {
                operation: "cumulative_returns",
                reason: "return series keys must be unique",
            });
        }
        if period != active_period {
            active_period.clone_from(&period);
            wealth = 1.0;
        }
        let value = values
            .get(index)
            .filter(|value| value.is_finite())
            .unwrap_or(0.0);
        wealth *= 1.0 + value;
        if !wealth.is_finite() {
            return Err(FerricAlphaError::InvalidInput {
                operation: "cumulative_returns",
                reason: "cumulative wealth must remain finite",
            });
        }
        output_dates.push(date);
        output_periods.push(period);
        output_wealth.push(wealth);
    }

    let mut columns = vec![date_column("date", output_dates, date.dtype())?];
    if has_period {
        columns.push(Series::new("period".into(), output_periods).into());
    }
    columns.push(Series::new("cumulative_return".into(), output_wealth).into());
    let frame = DataFrame::new(columns[0].len(), columns)?;
    finish_frame(frame, &sort)
}

fn validate_period_label(period: &str) -> Result<()> {
    let valid = period
        .strip_suffix('D')
        .filter(|number| !number.is_empty() && !number.starts_with('0'))
        .and_then(|number| number.parse::<u32>().ok())
        .is_some_and(|number| number > 0);
    if valid {
        Ok(())
    } else {
        Err(FerricAlphaError::InvalidInput {
            operation: "cumulative_returns",
            reason: "period must be a positive daily count",
        })
    }
}

pub fn positions(
    weights: &DataFrame,
    period: Period,
    sessions: Option<&DataFrame>,
) -> Result<DataFrame> {
    let date = required_column(weights, "date")?;
    let asset = required_column(weights, "asset")?;
    let weight = required_column(weights, "weight")?;
    if !matches!(date.dtype(), DataType::Datetime(_, _)) {
        return Err(FerricAlphaError::InvalidDtype {
            column: "date",
            expected: "Datetime",
            actual: date.dtype().to_string(),
        });
    }
    if !matches!(asset.dtype(), DataType::String) {
        return Err(FerricAlphaError::InvalidDtype {
            column: "asset",
            expected: "String",
            actual: asset.dtype().to_string(),
        });
    }
    if !matches!(weight.dtype(), DataType::Float64) {
        return Err(FerricAlphaError::InvalidDtype {
            column: "weight",
            expected: "Float64",
            actual: weight.dtype().to_string(),
        });
    }
    if date.null_count() > 0 || asset.null_count() > 0 {
        return Err(FerricAlphaError::NullKey);
    }

    let dates = date.as_materialized_series().datetime()?;
    let assets = asset.as_materialized_series().str()?;
    let values = weight.as_materialized_series().f64()?;
    let mut seen = HashSet::with_capacity(weights.height());
    let mut trades: BTreeMap<i64, BTreeMap<String, f64>> = BTreeMap::new();
    let mut asset_universe = BTreeSet::new();
    for ((date, asset), value) in dates
        .physical()
        .iter()
        .zip(assets.iter())
        .zip(values.iter())
    {
        let (Some(date), Some(asset), Some(value)) = (date, asset, value) else {
            return Err(FerricAlphaError::InvalidInput {
                operation: "positions",
                reason: "weights must be finite and non-null",
            });
        };
        if !value.is_finite() {
            return Err(FerricAlphaError::InvalidInput {
                operation: "positions",
                reason: "weights must be finite and non-null",
            });
        }
        if asset == "cash" {
            return Err(FerricAlphaError::InvalidInput {
                operation: "positions",
                reason: "asset name cash is reserved for synthetic cash",
            });
        }
        if !seen.insert((date, asset)) {
            return Err(FerricAlphaError::DuplicateKey);
        }
        asset_universe.insert(asset.to_owned());
        trades
            .entry(date)
            .or_default()
            .insert(asset.to_owned(), value);
    }

    let session_values = validate_sessions(sessions, date.dtype(), trades.keys().copied())?;
    let session_indexes = session_values
        .iter()
        .enumerate()
        .map(|(index, date)| (*date, index))
        .collect::<BTreeMap<_, _>>();
    if trades
        .keys()
        .any(|date| !session_indexes.contains_key(date))
    {
        return Err(FerricAlphaError::InvalidInput {
            operation: "positions",
            reason: "every trade date must be present in sessions",
        });
    }

    let assets = asset_universe.into_iter().collect::<Vec<_>>();
    let row_count = session_values.len() * (assets.len() + 1);
    let mut output_dates = Vec::with_capacity(row_count);
    let mut output_assets = Vec::with_capacity(row_count);
    let mut output_positions = Vec::with_capacity(row_count);
    let mut active = Vec::<ActiveSleeve>::new();

    for (session_index, session_date) in session_values.iter().enumerate() {
        active.retain(|sleeve| sleeve.expiry_index > session_index);
        if let Some(weights) = trades.get(session_date) {
            active.push(ActiveSleeve {
                expiry_index: session_index.saturating_add(period.count() as usize),
                weights: weights.clone(),
            });
        }

        let mut totals = BTreeMap::<String, f64>::new();
        for sleeve in &active {
            for (asset, weight) in &sleeve.weights {
                let total = totals.entry(asset.clone()).or_default();
                let updated = *total + weight;
                if !updated.is_finite() {
                    return Err(FerricAlphaError::InvalidInput {
                        operation: "positions",
                        reason: "active sleeve weights must sum to finite values",
                    });
                }
                *total = updated;
            }
        }
        let mut gross = 0.0;
        for value in totals.values() {
            gross += value.abs();
            if !gross.is_finite() {
                return Err(FerricAlphaError::InvalidInput {
                    operation: "positions",
                    reason: "gross exposure must remain finite",
                });
            }
        }
        let mut net = 0.0;
        for asset in &assets {
            let position = if gross > 0.0 {
                totals.get(asset).copied().unwrap_or(0.0) / gross
            } else {
                0.0
            };
            if !position.is_finite() {
                return Err(FerricAlphaError::InvalidInput {
                    operation: "positions",
                    reason: "normalized positions must remain finite",
                });
            }
            output_dates.push(*session_date);
            output_assets.push(asset.clone());
            output_positions.push(position);
            net += position;
            if !net.is_finite() {
                return Err(FerricAlphaError::InvalidInput {
                    operation: "positions",
                    reason: "net exposure must remain finite",
                });
            }
        }
        let cash = 1.0 - net;
        if !cash.is_finite() {
            return Err(FerricAlphaError::InvalidInput {
                operation: "positions",
                reason: "synthetic cash position must remain finite",
            });
        }
        output_dates.push(*session_date);
        output_assets.push("cash".to_owned());
        output_positions.push(cash);
    }

    let frame = DataFrame::new(
        output_dates.len(),
        vec![
            date_column("date", output_dates, date.dtype())?,
            Series::new("asset".into(), output_assets).into(),
            Series::new("position".into(), output_positions).into(),
        ],
    )?;
    finish_frame(frame, &["date", "asset"])
}

fn validate_sessions(
    sessions: Option<&DataFrame>,
    date_dtype: &DataType,
    inferred: impl Iterator<Item = i64>,
) -> Result<Vec<i64>> {
    let Some(sessions) = sessions else {
        return Ok(inferred.collect());
    };
    if sessions.width() != 1 {
        return Err(FerricAlphaError::InvalidInput {
            operation: "positions",
            reason: "sessions must contain exactly one date column",
        });
    }
    let date = required_column(sessions, "date")?;
    if !matches!(date.dtype(), DataType::Datetime(_, _)) {
        return Err(FerricAlphaError::InvalidDtype {
            column: "date",
            expected: "Datetime",
            actual: date.dtype().to_string(),
        });
    }
    if date.dtype() != date_dtype {
        return Err(FerricAlphaError::TimezoneMismatch);
    }
    if date.null_count() > 0 {
        return Err(FerricAlphaError::NullKey);
    }
    let mut values = date
        .as_materialized_series()
        .datetime()?
        .physical()
        .iter()
        .map(|value| value.expect("validated session date"))
        .collect::<Vec<_>>();
    values.sort_unstable();
    if values.windows(2).any(|window| window[0] == window[1]) {
        return Err(FerricAlphaError::DuplicateSession);
    }
    Ok(values)
}

pub fn factor_cumulative_returns(
    factor_data: &DataFrame,
    period: Period,
    options: PortfolioOptions,
) -> Result<DataFrame> {
    let filtered = filter_factor_data(factor_data, &options)?;
    ensure_requested_period(&filtered, period)?;
    let returns = factor_returns(
        &filtered,
        ReturnOptions::default().with_weight_options(options.weight_options),
    )?;
    let period_label = period.label();
    let periods = returns.column("period")?.as_materialized_series().str()?;
    let mask = BooleanChunked::from_iter_values(
        "period_filter".into(),
        periods.iter().map(|value| value == Some(&period_label)),
    );
    let selected = returns.filter(&mask)?;
    let adapted = DataFrame::new(
        selected.height(),
        vec![
            selected.column("date")?.clone(),
            selected
                .column("factor_return")?
                .clone()
                .with_name("return".into()),
        ],
    )?;
    cumulative_returns(&adapted)
}

pub fn factor_positions(
    factor_data: &DataFrame,
    period: Period,
    options: PortfolioOptions,
    sessions: Option<&DataFrame>,
) -> Result<DataFrame> {
    let filtered = filter_factor_data(factor_data, &options)?;
    ensure_requested_period(&filtered, period)?;
    let weights = factor_weights(&filtered, options.weight_options)?;
    positions(&weights, period, sessions)
}

fn ensure_requested_period(frame: &DataFrame, period: Period) -> Result<()> {
    let column = format!("forward_return_{}", period.label());
    if frame.column(&column).is_ok() {
        Ok(())
    } else {
        Err(FerricAlphaError::InvalidOption {
            option: "period",
            reason: "requested forward return period is not present",
        })
    }
}

pub(crate) fn filter_factor_data(
    frame: &DataFrame,
    options: &PortfolioOptions,
) -> Result<DataFrame> {
    let require_quantile = options.quantiles.is_some();
    let require_group = options.groups.is_some();
    crate::performance::frame::validate_factor_data(frame, require_quantile, require_group, true)?;

    let selected_quantiles = validate_quantile_filter(options.quantiles.as_deref())?;
    let selected_groups = validate_group_filter(options.groups.as_deref())?;
    let quantiles = frame
        .column("factor_quantile")
        .ok()
        .map(|column| column.as_materialized_series().u32())
        .transpose()?;
    let groups = frame
        .column("group")
        .ok()
        .map(|column| column.as_materialized_series().str())
        .transpose()?;
    let mask = BooleanChunked::from_iter_values(
        "portfolio_filter".into(),
        (0..frame.height()).map(|index| {
            let quantile_matches = selected_quantiles.as_ref().is_none_or(|selected| {
                quantiles
                    .as_ref()
                    .and_then(|quantiles| quantiles.get(index))
                    .is_some_and(|quantile| selected.contains(&quantile))
            });
            let group_matches = selected_groups.as_ref().is_none_or(|selected| {
                groups
                    .as_ref()
                    .and_then(|groups| groups.get(index))
                    .is_some_and(|group| selected.contains(group))
            });
            quantile_matches && group_matches
        }),
    );
    let filtered = frame.filter(&mask)?;
    if filtered.height() == 0 {
        return Err(FerricAlphaError::InsufficientData {
            operation: "portfolio filter",
        });
    }
    Ok(filtered)
}

fn validate_quantile_filter(values: Option<&[u32]>) -> Result<Option<HashSet<u32>>> {
    let Some(values) = values else {
        return Ok(None);
    };
    let selected = values.iter().copied().collect::<HashSet<_>>();
    if values.is_empty() || selected.len() != values.len() || selected.contains(&0) {
        return Err(FerricAlphaError::InvalidOption {
            option: "quantiles",
            reason: "quantiles must be positive and unique",
        });
    }
    Ok(Some(selected))
}

fn validate_group_filter(values: Option<&[String]>) -> Result<Option<HashSet<String>>> {
    let Some(values) = values else {
        return Ok(None);
    };
    let selected = values.iter().cloned().collect::<HashSet<_>>();
    if values.is_empty()
        || selected.len() != values.len()
        || selected.iter().any(|value| value.is_empty())
    {
        return Err(FerricAlphaError::InvalidOption {
            option: "groups",
            reason: "groups must be non-empty and unique",
        });
    }
    Ok(Some(selected))
}