finance-solution

finance-solution is a Rust library for time-value-of-money (TVM), cashflow, amortization, and related finance formulas — with detailed solution structs, period-by-period series, and pretty-printed tables. Scalar math executes in nanoseconds, making it suitable for quant systems. Solution structs execute in microseconds, providing observability with speed.

finance-solution is a financial library for solving time-value-of-money problems and financial math, including common quant calculations. 💸

People who will find this crate helpful include:

• Students of Finance who want to solve their financial problems using something better than a crude handheld calculator. Using this library also reduces the chance of human error, and provides better output and data displays than Excel in less time thereby providing a better surface for internalizing financial math concepts.
• New developers who want to learn Rust using Finance as a topic.
• Experienced Rust developers who want to learn more about finance and/or incorporate safe financial calculations into their projects.
• Serious Rust developers who want to build financial software, and prefer to rely on a rigourously tested library instead of reinventing the wheel and spending hundreds of hours to develop and test their own library of financial calculations.

In the v0.0.0 release, this library was geared only towards the basic financial equations, regarding:

• Simple Time-Value-of-Money formulas -- this includes present_value, future_value, rate, and periods (known as NPER in Excel).
• Cashflow Time-Value-of-Money formulas -- this includes present_value_annuity, future_value_annuity, net_present_value, and payment (known as PMT in Excel).
• Rate conversions -- this includes all conversions between apr, ear, and epr, and also includes conversion for continuous compounding (apr_continuous, ear_continuous).

Current Status: 0.2.0 (edition 2021, MSRV 1.70). Public math is Result-only (FinanceResult / FinanceError) — no dual panicking / try_* APIs. 0.1 established TVM, cashflow, amortization, returns, and stocks path metrics. 0.2 adds domain newtypes, full stocks::ta technical analysis (batch + incremental *State), teaching solutions/tables, and Criterion TA benches. The crate remains a math library to be consumed by other projects/engines/frameworks.

Modules (0.2 layout)

Quick start

use finance_solution::*;

// Present value of $4,000 in 3 years at 5% (simple compounding)
// Fixture rates/amounts are valid; production code should use `?` or match.
let pv = present_value(0.05, 3, 4_000.0, false).expect("fixture: 5% PV");
assert_rounded_2!(pv, -3_455.35); // sign: opposite of future value

// Prefer the solution API for formulas + period series
let answer = present_value_solution(0.05, 3, 4_000.0, false).expect("fixture");
println!("{}", answer.formula());
answer.series().print_table();

// Handle bad input without panicking
match present_value(-1.5, 3, 4_000.0, false) {
    Ok(v) => println!("pv = {v}"),
    Err(e) => eprintln!("error: {e}"),
}

Payment + amortization table

use finance_solution::*;

let rate = 0.08 / 12.0;
let periods = 60;
let principal = 13_000.0;

let solution = payment_solution(rate, periods, principal, 0.0, false).expect("fixture loan");
solution.print_table();

// Excel-style period components (1-based period index)
let interest_m1 = ipmt(rate, 1, periods, principal, 0.0, false).expect("period 1");
let principal_m1 = ppmt(rate, 1, periods, principal, 0.0, false).expect("period 1");
let year1_interest = cumipmt(rate, periods, principal, 0.0, 1, 12, false).expect("range");

Rate conversion

use finance_solution::*;

let r = apr(0.034, 12).expect("fixture: 3.4% APR monthly");
println!("EPR = {}, EAR = {}", r.epr(), r.ear());

Doubling rules & price paths

use finance_solution::*;

let d = doubling_solution(0.08).expect("fixture: 8%");
assert_rounded_2!(d.rule_of_72(), 9.0);
d.print_table();

let prices = [100.0, 110.0, 105.0, 120.0];
let path = price_path_solution(&prices, PricePathOptions::new(12.0)).expect("fixture prices");
path.print_summary();
path.series().print_table();

Technical analysis — quant pattern (stochastics)

finance-solution is a math library, not a trading engine: you own the bar loop and data feed. For indicators, define each variation once as a const parameter pack, validate once, then call .compute on every batch of bars (or free functions for one-off scripts).

use finance_solution::*;

// 1) Strategy knobs — fixed packs (FastStoch(9,3), FullStoch(14,3,3), …)
const FAST_9_3: StochasticParams = StochasticParams::fast(9, 3);
const FULL_14_3_3: StochasticParams = StochasticParams::full(14, 3, 3);
const MACD_STD: MacdParams = MacdParams::standard(); // (12, 26, 9)
const BB_20_2: BollingerParams = BollingerParams::standard();

// 2) Validate once at process/strategy startup (O(1))
//    Fallible constructors are named `new` → FinanceResult (not `try_new`):
//    fallibility lives in the return type, matching Result-only crate policy.
let stoch = ValidatedStochastic::new(FAST_9_3).expect("params");
let macd_eng = ValidatedMacd::new(MACD_STD).expect("params");
let bb = ValidatedBollinger::new(BB_20_2).expect("params");

// 3) Hot path — reuse engines on each symbol / day batch
// let kd = stoch.compute(&high, &low, &close)?;
// let m = macd_eng.compute(&closes)?;
// let bands = bb.compute(&closes)?;

// Teaching / audit path (formulas + print_table with n/a warm-up):
// let sol = stochastics_solution(&high, &low, &close, FAST_9_3)?;
// sol.print_table();

Common stochastic packs: StochasticParams::fast(9, 3), fast(14, 3), full(14, 3, 3), full(60, 10, 1). Same idea for MACD, Bollinger, Keltner, VWAP, RVOL — see stocks::ta rustdoc.

Run the verbose demo:

cargo run --example ta_indicators

Design notes

1. Result-only public math — domain failures are FinanceResult / FinanceError. There is no dual panicking + try_* pair; the ordinary name is the fallible function.
2. Solution + series — *_solution types carry inputs, symbolic/numeric formulas, and .series() for period detail.
3. Excel-compatible signs — loans/payments follow spreadsheet conventions (positive principal → negative payment, etc.).
4. Enums for flags — prefer [Compounding] and [PaymentTiming]; bool still converts via From for ergonomics. TA uses enums where modes matter (e.g. VwapMode, VwapPriceSource).
5. Tables — .print_table() / .print_table_locale() for teaching and debugging (copy/paste friendly).
6. Hot path vs solution path — scalar / series APIs are the production path (FinanceResult, no formula strings). Solution APIs are the teaching/observability path. Same formulas; different packaging. See benches/RESULTS.md.
7. TA params — one core function per indicator + Copy *Params + optional Validated*; presets via const constructors (fast, standard, …), not a zoo of nearly identical free functions.

Live bars: incremental state (not a market-data engine)

This crate does not subscribe to feeds or manage multi-symbol books. It does provide pure *State machines so your quant engine can update indicators on each payload (tick / 5s / 1m) without recomputing the full history:

Quant engine sketch (per-symbol pipeline; many symbols ⇒ HashMap<Symbol, SymbolPipeline>):

use finance_solution::*;

const FAST: StochasticParams = StochasticParams::fast(9, 3);
const MACD: MacdParams = MacdParams::standard();
const VWAP: VwapParams = VwapParams::cumulative_typical();

struct SymbolPipeline {
    stoch: StochState,
    macd: MacdState,
    vwap: VwapState,
    ema20: EmaState,
}

impl SymbolPipeline {
    fn new() -> FinanceResult<Self> {
        // *State::new is fallible (bad periods → Err), same idea as File::open
        Ok(Self {
            stoch: StochState::new(FAST)?,
            macd: MacdState::new(MACD)?,
            vwap: VwapState::new(VWAP)?,
            ema20: EmaState::new(20)?,
        })
    }

    /// Once per day/session (or on reconnect): seed from your store, then only push live bars.
    fn seed_history(
        &mut self,
        high: &[f64],
        low: &[f64],
        close: &[f64],
        volume: &[f64],
    ) -> FinanceResult<()> {
        self.stoch = StochState::from_history(FAST, high, low, close)?;
        self.macd = MacdState::from_history(MACD, close)?;
        self.vwap = VwapState::from_history(VWAP, high, low, close, volume)?;
        self.ema20 = EmaState::from_history(20, close)?;
        Ok(())
    }

    fn on_bar(&mut self, high: f64, low: f64, close: f64, volume: f64) -> FinanceResult<()> {
        let _kd = self.stoch.push(high, low, close)?;       // Option<(k,d)>
        let _m = self.macd.push(close)?;                    // Option<(macd,signal,hist)>
        let _vw = self.vwap.push(high, low, close, volume)?; // Option<f64>
        let _e = self.ema20.push(close)?;
        Ok(())
    }

    /// Multi-bar payload (e.g. several 1m bars in one message).
    fn on_bars(
        &mut self,
        high: &[f64],
        low: &[f64],
        close: &[f64],
        volume: &[f64],
    ) -> FinanceResult<()> {
        let _ = self.stoch.push_bars(high, low, close)?;
        let _ = self.macd.push_bars(close)?;
        let _ = self.vwap.push_bars(high, low, close, volume)?;
        let _ = self.ema20.push_bars(close)?;
        Ok(())
    }

    /// Your calendar decides this — the library never auto-resets at “market open”.
    fn on_session_open(&mut self) {
        self.vwap.reset(); // common: new day VWAP
        // stoch/macd/ema often CONTINUE across days; reset only if your strategy wants it
    }
}

Maximum control: continue multi-day series (never reset), or reset VWAP/RVOL at open, or rebuild state from a sliced history. All of that is your policy; we only expose the levers.

State is parity-tested against batch series (see stocks::ta::state tests). Also: SmaState, BollingerState, KeltnerState, RvolState.

Parameter order (when possible)

rate, periods, present_value, future_value, …

Money arguments often accept any Into<f64> (i32, f32, …). Rates stay f64; period counts are u32 (NPER returns f64 for fractional periods).

Error handling

Why. Libraries used from web handlers, batch jobs, or CLIs must not abort when a user types -150% interest or period 0.

Canonical pattern (structured match + ?):

use finance_solution::{payment, FinanceError, FinanceResult};

fn loan_payment(principal: f64) -> FinanceResult<f64> {
    payment(0.09 / 12.0, 60, principal, 0.0, false)
}

match loan_payment(20_000.0) {
    Ok(pmt) => println!("payment = {pmt}"),
    Err(FinanceError::InvalidRate { rate }) => {
        eprintln!("interest rate {rate} is not usable");
    }
    Err(FinanceError::NonFinite { field, value }) => {
        eprintln!("{field} must be finite; got {value}");
    }
    Err(e) => eprintln!("finance error: {e}"),
}

Affordability example (compose with ?):

use finance_solution::{payment, present_value, FinanceResult};

fn affordability(
    future_goal: f64,
    discount_rate: f64,
    loan_apr: f64,
    years: u32,
    loan_months: u32,
) -> FinanceResult<f64> {
    // Present value of the goal (Excel-style sign: positive FV → negative PV).
    let pv = present_value(discount_rate, years, future_goal, false)?;
    // Positive principal → negative payment (borrower cash out).
    let principal = pv.abs();
    payment(loan_apr / 12.0, loan_months, principal, 0.0, false)
}

// Call from a fallible main or handler:
// let pmt = affordability(50_000.0, 0.05, 0.06, 10, 60)?;
// pmt ≈ -593.43

Early payoff / mortgage what-ifs use the same Result APIs — see cargo run --example early_payoff_what_if and the amortization tables below.

Benefits:

1. No surprise panics on bad external input
2. Matchable variants for UI / metrics (FinanceError::code())
3. std::error::Error + Display for logging and ? into app error types
4. One name per formula — fallibility lives in the return type

Amortization: solution / series / tables

amortization_solution  →  .series()  →  .print_table()
       ↓                      ↓
   formulas + payment    period rows (principal, interest, balance, formulas)

use finance_solution::*;

let rate = 0.08 / 12.0;
let periods = 12;
let principal = 10_000.0;

let solution = amortization_solution(rate, periods, principal, 0.0, false)
    .expect("fixture: 12-month demo loan");
println!("{}", solution.formula());

let series = solution.series();
assert_eq!(series.len(), periods as usize);
series.print_table(true, true);

assert_approx_equal!(solution.ipmt(1).unwrap(), series[0].interest());
assert_approx_equal!(solution.ppmt(1).unwrap(), series[0].principal());
let _first_half = solution.cumipmt(1, 6).unwrap();

Sample print_table(true, true) (12-month $10k loan at 8% APR monthly):

period    payment  principal  interest     balance  principal_to_date  interest_to_date  payments_to_date  principal_remaining  interest_remaining  payments_remaining
------  ---------  ---------  --------  ----------  -----------------  ----------------  ----------------  -------------------  ------------------  ------------------
     1  -869.8843  -803.2176  -66.6667  9_196.7824          -803.2176          -66.6667         -869.8843          -9_196.7824           -371.9448         -9_568.7272
     2  -869.8843  -808.5724  -61.3119  8_388.2100        -1_611.7900         -127.9785       -1_739.7686          -8_388.2100           -310.6329         -8_698.8429
   ...
    12  -869.8843  -864.1235   -5.7608     -0.0000       -10_000.0000         -438.6115      -10_438.6115               0.0000             -0.0000              0.0000

Returns & stocks

use finance_solution::*;

let s = doubling_solution(0.08).expect("fixture");
s.print_table();
doubling_compare_rates(&[0.01, 0.05, 0.08, 0.12])
    .expect("fixture rates")
    .print_table();

let prices = [100.0, 102.0, 101.0, 108.0, 105.0, 112.0, 110.0, 118.0, 115.0, 125.0];
let path = price_path_solution(
    &prices,
    PricePathOptions::new(12.0).with_years(prices.len() as f64 / 12.0),
)
.expect("fixture prices");
path.print_summary();
path.series().print_table();

Examples

cargo run --example common_word_problems
cargo run --example amortization_table_demo
cargo run --example early_payoff_what_if
cargo run --example doubling_rules
cargo run --example price_path_analysis

Testing & CI

cargo test
cargo test --doc
cargo clippy --all-targets -- -D warnings
cargo bench   # optional Criterion suites (local / on demand)

Symmetry tests, Excel golden values, and integration tests cover core TVM, cashflow, amortization, returns, and stocks paths. CI runs build, tests, doctests, clippy, and fmt on stable and MSRV 1.70.

License

MIT — see LICENSE.