FA-62031 / Currency rounding and FX conversion / Open access
A partial close re-weights the average rate · case 01
After selling part of a long position the remaining lot shows a new average and wrong unrealized P&L.
ROOT CAUSE
The reduction branch recomputes avg as a signed weighted mean including the closing trade.
THE FAILURE
The reduction branch recomputes avg as a signed weighted mean including the closing trade.
Unsuccessful approach: Re-weighting with absolute quantities on reductions still moves the average of the surviving lot.
Case contract
solve(trades, spot): trades are [qty, rate] in foreign units (buy positive, sell negative) at home-per-foreign decimal rates. Adding in the position's direction updates the weighted average rate; reducing realizes closed_qty * (rate - avg) * sign(position) without changing avg; a flip realizes the closed part and opens the remainder at the trade rate; a flat position has no average. Zero-quantity trades are ignored. Unrealized = position * (spot - avg). Return [position, avg to 6 dp or None, realized to 2 dp, unrealized to 2 dp], rounded half-even only at output with unsigned zeros.
Why this case matters
Currency amounts must be rounded at the right stage and in the right unit, or ledgers, quotes and settlements drift by minor units.
1 / The failure
Exit 1"""Failure Map reference implementation. Python standard library only."""
import json
from decimal import Decimal, ROUND_HALF_EVEN, ROUND_HALF_UP, ROUND_FLOOR, ROUND_CEILING, ROUND_DOWN, ROUND_UP, ROUND_HALF_DOWN
from fractions import Fraction
N = 1
observations = []
def solve(trades, spot):
def fmt(fr, e):
d = (Decimal(fr.numerator) / Decimal(fr.denominator)).quantize(Decimal(1).scaleb(-e), rounding=ROUND_HALF_EVEN)
return format(abs(d) if d == 0 else d, 'f')
pos, avg, realized = 0, None, Fraction(0)
for qty, rate in trades:
r = Fraction(rate)
if qty == 0: continue
if pos == 0 or (pos > 0) == (qty > 0):
avg = r if pos == 0 else (abs(pos) * avg + abs(qty) * r) / (abs(pos) + abs(qty))
pos += qty
else:
closed = min(abs(qty), abs(pos))
sgn = 1 if pos > 0 else -1
realized += closed * (r - avg) * sgn
if pos + qty != 0: avg = (pos * avg + qty * r) / (pos + qty)
pos += qty
if pos == 0: avg = None
elif (pos > 0) != (sgn > 0): avg = r
unreal = pos * (Fraction(spot) - avg) if pos else Fraction(0)
return [pos, None if avg is None else fmt(avg, 6), fmt(realized, 2), fmt(unreal, 2)]
def check(label, actual, expected):
observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
cases = [[('regression reduction-average 1', ([[100, '1.10'], [100, '1.20'], [-50, '1.30']], '1.25'),
[150, '1.150000', '7.50', '15.00']),
('regression reduction-average 2', ([[300, '1.0711'], [-100, '1.0733'], [-100, '1.0745'], [-100, '1.0719']], '1.07'),
[0, None, '0.64', '0.00']),
('control: flip long to short', ([[100, '1.10'], [-150, '1.00']], '1.05'), [-50, '1.000000', '-10.00', '-2.50']),
('control: close short', ([[-100, '1.10'], [100, '1.00']], '1.05'), [0, None, '10.00', '0.00']),
('control: zero trade only', ([[0, '1.10']], '1.05'), [0, None, '0.00', '0.00']),
('control: short add', ([[-30, '1.0833'], [-70, '1.0851']], '1.0800'), [-100, '1.084560', '0.00', '0.46'])],
[('regression reduction-average 1',
([[340, '1.0922'], [380, '1.0595'], [0, '1.1288'], [220, '1.0703'], [220, '1.0901'], [-390, '1.1478']], '1.0752'),
[770, '1.076936', '27.64', '-1.34']),
('regression reduction-average 2', ([[270, '1.1098'], [-40, '1.1476']], '1.0552'),
[230, '1.109800', '1.51', '-12.56']),
('control: zero trade only', ([[0, '1.10']], '1.05'), [0, None, '0.00', '0.00']),
('control: short add', ([[-30, '1.0833'], [-70, '1.0851']], '1.0800'), [-100, '1.084560', '0.00', '0.46']),
('control: three reductions', ([[300, '1.0711'], [-100, '1.0733'], [-100, '1.0745'], [-100, '1.0719']], '1.07'),
[0, None, '0.64', '0.00']),
('control: zero between', ([[50, '1.1'], [-50, '1.2'], [0, '1.3'], [20, '1.4']], '1.5'),
[20, '1.400000', '5.00', '2.00'])],
[('regression reduction-average 1',
([[-540, '1.1190'], [-60, '1.0677'], [20, '1.1443'], [-240, '1.1212'], [0, '1.0908']], '1.0791'),
[-820, '1.116015', '-0.61', '30.27']),
('regression reduction-average 2',
([[-450, '1.1009'], [0, '1.1001'], [230, '1.0743'], [0, '1.0884'], [-190, '1.0956'], [30, '1.0635']], '1.0627'),
[-380, '1.098444', '7.17', '13.58']),
('control: zero between', ([[50, '1.1'], [-50, '1.2'], [0, '1.3'], [20, '1.4']], '1.5'),
[20, '1.400000', '5.00', '2.00']),
('control: flip short to long', ([[-40, '1.2'], [100, '1.1']], '1.15'), [60, '1.100000', '4.00', '3.00']),
('control: add then reduce', ([[100, '1.10'], [100, '1.20'], [-50, '1.30']], '1.25'),
[150, '1.150000', '7.50', '15.00']),
('control: flip long to short', ([[100, '1.10'], [-150, '1.00']], '1.05'), [-50, '1.000000', '-10.00', '-2.50'])],
[('regression reduction-average 1',
([[-40, '1.1445'], [-570, '1.0711'], [530, '1.1387'], [-450, '1.0688'], [-300, '1.1071']], '1.1211'),
[-830, '1.083329', '-33.28', '-31.35']),
('regression reduction-average 2',
([[-460, '1.1380'], [-260, '1.0732'], [-130, '1.1234'], [-300, '1.0815'], [210, '1.0636']], '1.0880'),
[-940, '1.106960', '9.11', '17.82']),
('control: flip long to short', ([[100, '1.10'], [-150, '1.00']], '1.05'), [-50, '1.000000', '-10.00', '-2.50']),
('control: close short', ([[-100, '1.10'], [100, '1.00']], '1.05'), [0, None, '10.00', '0.00']),
('control: zero trade only', ([[0, '1.10']], '1.05'), [0, None, '0.00', '0.00']),
('control: short add', ([[-30, '1.0833'], [-70, '1.0851']], '1.0800'), [-100, '1.084560', '0.00', '0.46'])],
[('regression reduction-average 1',
([[0, '1.0935'], [100, '1.0751'], [570, '1.0814'], [280, '1.1411'], [-350, '1.1101'], [590, '1.1297']], '1.0775'),
[1190, '1.113885', '4.12', '-43.30']),
('regression reduction-average 2',
([[0, '1.0983'], [-530, '1.1416'], [-170, '1.1049'], [280, '1.1333'], [-140, '1.0910'], [0, '1.0954']], '1.0893'),
[-560, '1.122265', '-0.17', '18.46']),
('control: short add', ([[-30, '1.0833'], [-70, '1.0851']], '1.0800'), [-100, '1.084560', '0.00', '0.46']),
('control: three reductions', ([[300, '1.0711'], [-100, '1.0733'], [-100, '1.0745'], [-100, '1.0719']], '1.07'),
[0, None, '0.64', '0.00']),
('control: zero between', ([[50, '1.1'], [-50, '1.2'], [0, '1.3'], [20, '1.4']], '1.5'),
[20, '1.400000', '5.00', '2.00']),
('control: flip short to long', ([[-40, '1.2'], [100, '1.1']], '1.15'), [60, '1.100000', '4.00', '3.00'])]]
for label, args, expected in cases[N - 1]:
check(label, solve(*args), expected)
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)
| Boundary fixture | Actual | Expected | Outcome |
|---|---|---|---|
| regression reduction-average 1 | [150, '1.100000', '7.50', '22.50'] | [150, '1.150000', '7.50', '15.00'] | Failed |
| regression reduction-average 2 | [0, None, '1.31', '0.00'] | [0, None, '0.64', '0.00'] | Failed |
| control: flip long to short | [-50, '1.000000', '-10.00', '-2.50'] | [-50, '1.000000', '-10.00', '-2.50'] | Passed |
| control: close short | [0, None, '10.00', '0.00'] | [0, None, '10.00', '0.00'] | Passed |
| control: zero trade only | [0, None, '0.00', '0.00'] | [0, None, '0.00', '0.00'] | Passed |
| control: short add | [-100, '1.084560', '0.00', '0.46'] | [-100, '1.084560', '0.00', '0.46'] | Passed |
SHA-256 / 6f0ed5f4eb79a0f4c78965a7c9ab40f352b6e4321ba1181686f797280efe2a81
2 / The unsuccessful fix
Exit 1"""Failure Map reference implementation. Python standard library only."""
import json
from decimal import Decimal, ROUND_HALF_EVEN, ROUND_HALF_UP, ROUND_FLOOR, ROUND_CEILING, ROUND_DOWN, ROUND_UP, ROUND_HALF_DOWN
from fractions import Fraction
N = 1
observations = []
def solve(trades, spot):
def fmt(fr, e):
d = (Decimal(fr.numerator) / Decimal(fr.denominator)).quantize(Decimal(1).scaleb(-e), rounding=ROUND_HALF_EVEN)
return format(abs(d) if d == 0 else d, 'f')
pos, avg, realized = 0, None, Fraction(0)
for qty, rate in trades:
r = Fraction(rate)
if qty == 0: continue
if pos == 0 or (pos > 0) == (qty > 0):
avg = r if pos == 0 else (abs(pos) * avg + abs(qty) * r) / (abs(pos) + abs(qty))
pos += qty
else:
closed = min(abs(qty), abs(pos))
sgn = 1 if pos > 0 else -1
realized += closed * (r - avg) * sgn
if pos + qty != 0: avg = (abs(pos) * avg + abs(qty) * r) / (abs(pos) + abs(qty))
pos += qty
if pos == 0: avg = None
elif (pos > 0) != (sgn > 0): avg = r
unreal = pos * (Fraction(spot) - avg) if pos else Fraction(0)
return [pos, None if avg is None else fmt(avg, 6), fmt(realized, 2), fmt(unreal, 2)]
def check(label, actual, expected):
observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
cases = [[('regression reduction-average 1', ([[100, '1.10'], [100, '1.20'], [-50, '1.30']], '1.25'),
[150, '1.150000', '7.50', '15.00']),
('regression reduction-average 2', ([[300, '1.0711'], [-100, '1.0733'], [-100, '1.0745'], [-100, '1.0719']], '1.07'),
[0, None, '0.64', '0.00']),
('control: flip long to short', ([[100, '1.10'], [-150, '1.00']], '1.05'), [-50, '1.000000', '-10.00', '-2.50']),
('control: close short', ([[-100, '1.10'], [100, '1.00']], '1.05'), [0, None, '10.00', '0.00']),
('control: zero trade only', ([[0, '1.10']], '1.05'), [0, None, '0.00', '0.00']),
('control: short add', ([[-30, '1.0833'], [-70, '1.0851']], '1.0800'), [-100, '1.084560', '0.00', '0.46'])],
[('regression reduction-average 1',
([[340, '1.0922'], [380, '1.0595'], [0, '1.1288'], [220, '1.0703'], [220, '1.0901'], [-390, '1.1478']], '1.0752'),
[770, '1.076936', '27.64', '-1.34']),
('regression reduction-average 2', ([[270, '1.1098'], [-40, '1.1476']], '1.0552'),
[230, '1.109800', '1.51', '-12.56']),
('control: zero trade only', ([[0, '1.10']], '1.05'), [0, None, '0.00', '0.00']),
('control: short add', ([[-30, '1.0833'], [-70, '1.0851']], '1.0800'), [-100, '1.084560', '0.00', '0.46']),
('control: three reductions', ([[300, '1.0711'], [-100, '1.0733'], [-100, '1.0745'], [-100, '1.0719']], '1.07'),
[0, None, '0.64', '0.00']),
('control: zero between', ([[50, '1.1'], [-50, '1.2'], [0, '1.3'], [20, '1.4']], '1.5'),
[20, '1.400000', '5.00', '2.00'])],
[('regression reduction-average 1',
([[-540, '1.1190'], [-60, '1.0677'], [20, '1.1443'], [-240, '1.1212'], [0, '1.0908']], '1.0791'),
[-820, '1.116015', '-0.61', '30.27']),
('regression reduction-average 2',
([[-450, '1.1009'], [0, '1.1001'], [230, '1.0743'], [0, '1.0884'], [-190, '1.0956'], [30, '1.0635']], '1.0627'),
[-380, '1.098444', '7.17', '13.58']),
('control: zero between', ([[50, '1.1'], [-50, '1.2'], [0, '1.3'], [20, '1.4']], '1.5'),
[20, '1.400000', '5.00', '2.00']),
('control: flip short to long', ([[-40, '1.2'], [100, '1.1']], '1.15'), [60, '1.100000', '4.00', '3.00']),
('control: add then reduce', ([[100, '1.10'], [100, '1.20'], [-50, '1.30']], '1.25'),
[150, '1.150000', '7.50', '15.00']),
('control: flip long to short', ([[100, '1.10'], [-150, '1.00']], '1.05'), [-50, '1.000000', '-10.00', '-2.50'])],
[('regression reduction-average 1',
([[-40, '1.1445'], [-570, '1.0711'], [530, '1.1387'], [-450, '1.0688'], [-300, '1.1071']], '1.1211'),
[-830, '1.083329', '-33.28', '-31.35']),
('regression reduction-average 2',
([[-460, '1.1380'], [-260, '1.0732'], [-130, '1.1234'], [-300, '1.0815'], [210, '1.0636']], '1.0880'),
[-940, '1.106960', '9.11', '17.82']),
('control: flip long to short', ([[100, '1.10'], [-150, '1.00']], '1.05'), [-50, '1.000000', '-10.00', '-2.50']),
('control: close short', ([[-100, '1.10'], [100, '1.00']], '1.05'), [0, None, '10.00', '0.00']),
('control: zero trade only', ([[0, '1.10']], '1.05'), [0, None, '0.00', '0.00']),
('control: short add', ([[-30, '1.0833'], [-70, '1.0851']], '1.0800'), [-100, '1.084560', '0.00', '0.46'])],
[('regression reduction-average 1',
([[0, '1.0935'], [100, '1.0751'], [570, '1.0814'], [280, '1.1411'], [-350, '1.1101'], [590, '1.1297']], '1.0775'),
[1190, '1.113885', '4.12', '-43.30']),
('regression reduction-average 2',
([[0, '1.0983'], [-530, '1.1416'], [-170, '1.1049'], [280, '1.1333'], [-140, '1.0910'], [0, '1.0954']], '1.0893'),
[-560, '1.122265', '-0.17', '18.46']),
('control: short add', ([[-30, '1.0833'], [-70, '1.0851']], '1.0800'), [-100, '1.084560', '0.00', '0.46']),
('control: three reductions', ([[300, '1.0711'], [-100, '1.0733'], [-100, '1.0745'], [-100, '1.0719']], '1.07'),
[0, None, '0.64', '0.00']),
('control: zero between', ([[50, '1.1'], [-50, '1.2'], [0, '1.3'], [20, '1.4']], '1.5'),
[20, '1.400000', '5.00', '2.00']),
('control: flip short to long', ([[-40, '1.2'], [100, '1.1']], '1.15'), [60, '1.100000', '4.00', '3.00'])]]
for label, args, expected in cases[N - 1]:
check(label, solve(*args), expected)
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)
| Boundary fixture | Actual | Expected | Outcome |
|---|---|---|---|
| regression reduction-average 1 | [150, '1.180000', '7.50', '10.50'] | [150, '1.150000', '7.50', '15.00'] | Failed |
| regression reduction-average 2 | [0, None, '0.44', '0.00'] | [0, None, '0.64', '0.00'] | Failed |
| control: flip long to short | [-50, '1.000000', '-10.00', '-2.50'] | [-50, '1.000000', '-10.00', '-2.50'] | Passed |
| control: close short | [0, None, '10.00', '0.00'] | [0, None, '10.00', '0.00'] | Passed |
| control: zero trade only | [0, None, '0.00', '0.00'] | [0, None, '0.00', '0.00'] | Passed |
| control: short add | [-100, '1.084560', '0.00', '0.46'] | [-100, '1.084560', '0.00', '0.46'] | Passed |
SHA-256 / e97bfaf21d969d87aea0ec417b57cd2be86d0dd134b4e0b4f52cc70aeaafcea1
HELD IN THE MEMBER ARCHIVE
The verified repair and its recorded checks are member-only.
This mechanism has 6 recorded checks per implementation. The open-access tier publishes the failure and the unsuccessful fix; the repaired source that passes every check, and the observations that prove it, are available to members.
Every case sharing this mechanism uses the same contract and the same repair, so this one record is held back for all of them.
Member access is invitation-based. Sign in with your invited account to inspect the repair.
Sign in to the archive ↗Verification & scope
A deterministic, bounded teaching model with a stipulated toy contract; it makes no claim of conformance to any real regulation, standard, or institution's rules. This reproducer isolates one failure mechanism. Results cover the supplied fixtures. Variants within a family share a test contract and should remain grouped when constructing evaluation splits. Related mechanisms with a shared evaluation_group must also remain together; these controlled models are not independent production incidents.
Observations recorded using Python 3.12.14 at 2026-09-29T14:47:00.757985+00:00.
Case digest / f95daabf5da8cfb2af9e19be9b32e7751e69bcfa72387ca73ceaa95250a59f89