Skip to main content

Multi-asset rebalance

Problem: You have separate strategies for equities, crypto, and fx. You want to allocate capital across them and rebalance the composite weekly.

Solution: Sleeve composition via mts1b-portfolio.multi_sleeve.

The setup​

Three sleeves, each producing target weights internally:

sleeves = {
    "equities_momentum": {
        "factor": "f_momentum_12_1",
        "universe": "russell-1000",
        "rebal": "monthly",
        "allocation": 0.5,        # 50% of fund NAV
    },
    "crypto_mean_rev": {
        "factor": "f_crypto_realized_vol",
        "universe": "crypto-top-10",
        "rebal": "weekly",
        "allocation": 0.3,
    },
    "fx_carry": {
        "factor": "f_fx_carry",
        "universe": "g10-fx",
        "rebal": "monthly",
        "allocation": 0.2,
    },
}

Composition​

from mts1b_portfolio.multi_sleeve import compose
from mts1b_oms import OmsClient

oms = OmsClient(fund_id="multi-asset-paper", base_url="http://oms:8001")

# Each rebalance window, run all sleeves, compose, diff against current
async def rebalance(asof):
    sleeve_signals = {}
    for name, cfg in sleeves.items():
        signal = await mts1b_research.run_sleeve(name, asof=asof)
        sleeve_signals[name] = {
            "weights": signal.weights,
            "allocation": cfg["allocation"],
        }

    composite = compose(sleeves=sleeve_signals, overlay="hrp")
    # composite is dict[symbol → target_weight] summing to 1.0 (gross)

    current = await oms.get_positions()
    diffs = compute_diff(current, composite, nav=await oms.get_nav())

    for trade in diffs:
        await oms.submit(trade.to_order())

Sleeve overlap handling​

If a symbol appears in two sleeves (e.g. SPY in both equities momentum and a hedging sleeve), compose sums the contributions:

sleeves = {
    "equities_long":  {"weights": {"SPY": 0.4, "QQQ": 0.6}, "allocation": 0.6},
    "spy_hedge":      {"weights": {"SPY": -0.5},             "allocation": 0.4},
}

composite = compose(sleeves=sleeves, overlay="equal")
# {"SPY": 0.6*0.4 + 0.4*(-0.5) = 0.04,
#  "QQQ": 0.6*0.6 = 0.36}

Net SPY exposure is small (mostly hedged); QQQ exposure is positive. The composer hides this complexity from each sleeve.

HRP overlay on sleeve returns​

Instead of fixed allocation, derive from each sleeve's historical returns:

composite = compose(
    sleeves=sleeve_signals,
    overlay="hrp",                   # HRP on sleeve returns matrix
    hrp_lookback_days=252,
)

If equities and crypto are negatively correlated over the last year, HRP gives them more allocation (diversification). If they've recently correlated, less.

Turnover budget​

Aggressive composition causes high turnover at the seams (sleeve A says "buy SPY", sleeve B says "sell SPY"). Constrain it:

composite = compose(
    sleeves=sleeve_signals,
    overlay="hrp",
    max_weekly_turnover=0.30,        # max 30% of NAV traded per week
)

The composer solves a constrained problem: minimize tracking error to target while respecting the turnover cap.

Capacity gating​

If the equities momentum sleeve wants to buy $50k of a $20m ADV stock and your envelope says max_position_pct=1% of ADV, the order is capped:

from mts1b_portfolio.capacity import position_capacity

cap = position_capacity(symbol="TINY_CORP", nav_usd=fund_nav, fraction_of_adv=0.01)
order_qty = min(target_qty, cap / order.limit_price)

mts1b-riskengine enforces this as a hard gate. The composer respects it proactively so the order isn't rejected.

Reporting​

After rebalance:

from mts1b_reportslibrary import sleeve_report

report = await sleeve_report(
    fund_id="multi-asset-paper",
    sleeves=list(sleeves.keys()),
    window="ytd",
)
# Per-sleeve P/L, Sharpe, contribution to fund return, turnover
print(report.to_table())

See also​